article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in higher dimensions a true crystal , named wigner crystal , where the density is perfectly ordered , is possible when the repulsion between particles dominates over the kinetic energy in the case of small density . in one dimension , quantum fluctuations prevent any type of true spin - symmetry breaking and long - range order in the ground state , and thus the crystal does not have perfect order @xcite .
however , in one dimension , the equilibrium positions of the electrons are assumed to be equally spaced and a solid like state can be produced when the correlation length exceeds the sample length @xcite .
the liquid - to - quasi - wigner crystal crossover when the density decreasing can be characterized by the onset of the @xmath1-peak in the static structure factor following from the luttinger liquid predictions @xcite , with @xmath2 the fermi wave vector ( related to the fermion density @xmath4 by @xmath5 ) , or by the @xmath6 high - contrast peaks in the pair correlation function @xcite , with @xmath3 the particle number in the system . in the system where the constituent particles are not electrons but dilute atoms
, the fundamental interaction is no longer of long - range coulomb type but of short - range contact one , which is the case in the field of ultracold atomic systems . in this system
, the ground - state density profile displays at strong coupling well - pronounced wigner oscillations with a ` @xmath7 periodicity '
. however , the static structure factor or the pair correlation function does not show any signature of wigner - molecule - type correlations like that in the system of long - range interaction @xcite , which means wigner - type molecules are absent in the system of short - range interaction @xcite . in this paper
, we focus only on the @xmath8 crossover " in the density profiles : for vanishing and weak interactions the ground state is liquid - like with @xmath9-friedel oscillations , whereas at strong repulsions a periodicity of @xmath10-wigner oscillation emerges . for the one - dimensional ( 1d ) harmonically trapped system of spin - balanced or imbalanced fermionic atomic gases ,
many numerical schemes have been used in studying lots of interesting quantum effects @xcite . a density - functional scheme using as reference the bethe - ansatz solvable homogeneous system was proposed to calculate the ground - state properties @xcite .
however , this approach was found to fail at the strong - coupling limit @xcite .
more precisely , the exchange - correlation potential proposed there is not able to describe the @xmath11 crossover , which , as noted above , is expected to occur upon increasing the strength of the repulsive interactions between antiparallel - spin particles .
a simple functional is proposed in ref .
[ ] for a two - particle system that embodies this crossover and is capable of describing inhomogeneous luttinger liquids with strong repulsions .
the main idea is to capture the tendency to @xmath1-wigner oscillations by adding an infinitesimal spin - symmetry - breaking magnetic field to the hamiltonian and by resorting to spin - density functional theory ( sdft ) based on the spin - polarized bethe - ansatz equations @xcite . for fermions loaded in a 1d optical lattice
, the system can be modeled by a 1d fermi - hubbard model .
friedel oscillations of wavelength @xmath9 and @xmath7 component are analyzed in this model with open boundary condition by both the bethe - ansatz based sdft , the exact diagonalization @xcite , bosonization , and numerical density matrix renormalization group calculations @xcite .
it was found that the @xmath11 crossover takes place at @xmath12 .
for an unpolarized system , if a tiny magnetization which is proportional to the difference of the spin - up and spin - down densities is induced by hand into the exchange - correlation potential , the @xmath1-oscillation appears in the density profile at the expense of violating the lieb - mattis theorem , while for a polarized system , sdft can reproduce both the @xmath9 and @xmath7 oscillations without violating the lieb - mattis theorem @xcite .
it is found that the @xmath1 density oscillations is strongly related to the large fermi sea induced by the strong repulsion between different spin species @xcite .
some 1d systems , for example , the 1d gaudin - yang gas in a homogeneous continuum space , or the 1d lieb - wu model of particles in a discrete lattice , i.e. , 1d fermi - hubbard model , where the fermions interact through attractive or repulsive short - range interactions , are exactly solvable .
however , in the inhomogeneous case only some special systems can be solved exactly .
for example , the 1d two - component fermi gas in an infinite potential well can be solved exactly by the bethe - ansatz techniques , where the ground state properties are investigated @xcite .
the density profiles are evaluated from weak to very strong interactions . in the strongly interacting limit
, the density profile tends to that of the spinless fermions though the momentum distributions differ .
an exact analytical solution of 1d spin-1/2 fermions with infinite repulsion for arbitrary confining potential is obtained by the combination of girardeau s hard - core contacting boundary condition and group theoretical method @xcite . for the spin - imbalanced case they found that the spin - up and spin - down density distributions show alternative peaks avoiding overlapping together and the parity symmetry . in this paper
, we will employ the bethe - ansatz based sdft to investigate the 1d inhomogeneous system with general interaction strengths .
our results are compared against the above mentioned exact analytic and the exact numerical diagonalization method .
the wigner oscillations while increasing coupling strength for both unpolarized and polarized cases will be studied .
the rest of the paper is organized as follows . in sec .
ii , we present the model and the method .
the numerical results are followed in sec .
iii . in sec .
iv , we conclude what we found .
in this work we consider a 1d system of two - component fermi gas with a contact repulsive interaction of strength @xmath13
the two fermion species are taken to have the same mass @xmath14 , with @xmath15 spin - up species and @xmath16 spin - down ones .
the total number of particles is @xmath17 .
the system is described by the hamiltonian , @xmath18 where contact interactions between parallel - spin particles are suppressed by the pauli exclusion principle and @xmath19 the confining potential .
we consider that the atoms are either confined in a 1d box of length @xmath20 ( @xmath21 inside the box and @xmath22 elsewhere ) or subjected to a strongly anisotropic harmonic potential @xmath23 , characterized by angular frequencies @xmath24 and @xmath25 in the radial and axial directions with @xmath26 . for the system of harmonic trap , we choose the harmonic - oscillator length @xmath27 as unit of length and the harmonic - oscillator quantum @xmath28 as unit of energy .
the hamiltonian ( [ eq : igy ] ) can be shown to be governed by the dimensionless coupling parameter @xmath29 for @xmath21 the hamiltonian ( [ eq : igy ] ) reduces to the homogeneous gaudin - yang model solvable by means of the bethe - ansatz technique @xcite and determined by the linear density @xmath30 , by the spin polarization @xmath31 , and by the interaction strength @xmath13 . for a balanced fermi gas with @xmath32 confined in a harmonic trap for various interaction strengths .
the crossover from @xmath0- to @xmath1-oscillations with increasing interaction strength @xmath33 is showed by the doubling of the crests in the density profile .
the dominance of the @xmath1-oscillations over @xmath0-ones can be better evidenced by the density derivative @xmath34 . in the inset ,
the derivative of the density with respect to @xmath35 is shown for @xmath36 , and @xmath37 as an indicator to examine the @xmath11 crossover .
[ fig : one ] ] unit ) as a function of @xmath35 ( in arbitrary length unit ) for a polarized fermi gas with @xmath38 and @xmath39 confined in a box of length @xmath40 for various interaction strengths .
the sdft results are plotted in symbols and the exact bethe - ansatz ones in ( colored ) lines .
the thin ( black ) solid line represents the result in the infinite repulsive interaction limit .
[ fig : three ] ] the ground - state spin - density distributions @xmath41 can be calculated within sdft by solving self - consistently the kohn - sham ( ks ) equation , @xmath42(z)\right]\varphi_{\alpha,\sigma}(z)=\varepsilon_{\alpha,\sigma}\varphi_{\alpha,\sigma}(z)\ ] ] together with the closure @xmath43 here , @xmath44(z)=v^{(\sigma)}_{\rm
xc}[n_\sigma](z)+v_{\rm ext}(z)$ ] is the spin - dependent effective ks potential .
the first term in @xmath45 is the mean - field term @xmath46 , while the second term is the exchange - correlation potential defined as the functional derivative of the exchange - correlation energy @xmath47 $ ] evaluated at the ground - state density profile , @xmath48/\delta n_\sigma(z)|_{\rm \scriptscriptstyle gs}$ ] . to obtain the density distribution @xmath41 from eqs .
( [ eq : kss ] ) and ( [ eq : closure ] ) , approximations to the exchange - correlation functions @xmath49 are unavoidable . the local - spin - density approximation ( lsda )
is known to provide an excellent description of the ground - state properties of a variety of inhomogeneous systems @xcite . in the following
we employ a bethe - ansatz - based lsda ( balsda ) functional for the exchange - correlation potential , @xmath50 & \rightarrow & e^{\rm lsda}_{\rm xc}[n_\sigma]\\ & = & \int dz\ , n(z)\left.\varepsilon^{\rm hom}_{\rm xc}(n,\zeta , g_{\rm 1d})\right|_{n\rightarrow n(z),\zeta\rightarrow \zeta(z)}\,,\nonumber\end{aligned}\ ] ] where the exchange - correlation energy @xmath51 of the homogeneous gaudin - yang model is defined by @xmath52 here @xmath53 is the ground - state energy of the exact bethe - ansatz solution of the model , @xmath54 is the mean - field energy , and @xmath55 is the noninteracting kinetic energy @xcite .
= 0 cm [ cols="^,^ " , ] = 0 cm and @xmath56 for a polarized fermi gas with @xmath38 and @xmath39 confined in a harmonic trap .
the results are obtained based on the sdft .
[ fig : five],title="fig : " ]
in this section , we present our numerical results by employing the bethe - ansatz based sdft to investigate the 1d inhomogeneous system with broad interaction strengths . our results are compared against the exact numerical diagonalization method while this method is available when the system is small . for the polarized system in a box
, we compare our results with the analytical ones which are available in this case by the exact bethe - ansatz method . in fig .
[ fig : one ] , we study an unpolarized fermi gas with the number of total particles @xmath57 confined in a harmonic trap .
the ground - state density distributions is calculated for various interaction strengths within sdft . for the unpolarized system ,
the ks equations have to be solved self - consistently by taking one of the following methods to obtain the correct @xmath11 crossover .
( i ) one can either add an infinitesimal spin - symmetry - breaking magnetic field .
the field strength is infinitesimal small and reduced to zero during the self - consistent solution of the ks equations .
( ii ) alternatively one can solve the ks equations from an initial guess which has slightly broken spin symmetry [ _ i.e. _ an initial guess with @xmath58 , without using any @xmath59 field , ( iii ) the last possibility is to break the symmetries of the density distributions just one time by shifting the center of the densities during the self - consistent cycle [ for example , let @xmath60 with @xmath61 a nonzero value ] .
when @xmath33 is small , the above mentioned methods restore the results from the bethe - ansatz based local density approximation ( balda ) @xcite .
however when @xmath33 is large , the final result for the total density @xmath62 from the balsda is very different from that resulting from the balda theory @xcite .
we want to mention here that , without using the above mentioned symmetry - breaking procedure , the balsda theory produces the same results as that of balda .
that is , for the balsda theory , to achieve a correct shape for the density profile , one has to use one of the ( i)-(iii ) procedure @xcite .
this method has long been used in the literature in discussing the ground - state of the @xmath63 molecule at any bond length .
the symmetry - breaking choice typically produce a spin - unpolarized density out to a critical bond length by localizing an up - spin electron on one nucleus and a down - spin electron on the other . while this leads to an incorrect spin - density , but it leads to the correct dissociation limit for the energy and the correct total density @xcite . at the weak interaction regime , the density is characterized by @xmath0-friedel oscillations with @xmath64 peaks . as the interaction strength increases to the strongly interacting limit , the number of peaks in the density profile is doubled .
the atoms are distributed in order to minimize the interparticle repulsion energy .
this leads to the emergence of @xmath3 crests in the density profile characterized by strong correlations and the @xmath1-wigner oscillations . in this case
the atoms behave like a fully polarized ( or noninteracting spinless ) fermions as in the tonks - girardeau gas @xcite .
numerically we find that a crossover from the @xmath0-friedel oscillations of @xmath64 peaks to the @xmath1-wigner oscillations happens around @xmath65 accompanied of @xmath3 peaks .
the crossover is smooth , see , for example , the extensive density - matrix renormalization group study by sffing _ et al .
_ @xcite . in the inset ,
a derivative of the density with respect to @xmath66 is used as an indicator to signal the @xmath11 crossover .
we find more clearly that @xmath3 peaks appear after the threshold value . and
@xmath56 for a polarized fermi gas with @xmath67 and @xmath68 confined in a harmonic trap .
the results are obtained based on the sdft .
[ fig : six ] ] and @xmath56 for a polarized fermi gas with @xmath69 and @xmath70 confined in a harmonic trap .
the results are obtained based on the sdft .
[ fig : eight ] ] and @xmath56 for a polarized fermi gas with @xmath71 and @xmath72 confined in a harmonic trap .
the results are obtained based on the sdft .
[ fig : nine ] ] in order to evaluate the credibility of the present method , we compare our results in the following against exact methods for a polarized system in a box ( in [ sec1 ] ) and a harmonically confined system of small number of fermi atoms ( in [ sec2 ] ) .
in this section we consider a polarized fermi gas with @xmath15 up - spin fermions and @xmath16 down - spin fermions confined in a box of length @xmath20 , which is exactly solvable by the bethe - ansatz method @xcite . in fig .
[ fig : three ] , we illustrate the ground - state density distributions as a function of @xmath35 based on the bethe - ansatz based sdft for @xmath38 and @xmath39 .
@xmath20 is chosen as units of length .
the interaction strength sets the relative importance of the kinetic energy and the interaction . for weak interaction strength ,
@xmath73 peaks appear in the density .
when interaction strength becomes stronger ( around @xmath74 ) , @xmath75 peaks emerge . at even stronger interaction strength
, exchange is suppressed and the atoms act as though they were noninteracting and spinless in analogy to the case studied for bosons with an infinite contact repulsion @xcite . for @xmath76 , the infinite repulsive interaction limit is reached , which is compared to the spinless fermions results . in general , when the interaction strength is large enough , the density @xmath62 distributes between @xmath3 maxima , which is different from the confined 1d electrons of long - range coulomb interactions where the density vanishes almost completely between well separated maxima @xcite .
we find , in almost all the range of the interaction strength , sdft gives good results comparing against the exact bethe - ansatz ones .
the good agreement may be that the exact exchange - correlation energy of the homogeneous system has almost the same form as that of the inhomogeneous ( hard - wall ) one which is also integrable . here
we report our studies on the ground - state density distributions for a polarized fermi gas confined in a harmonic trap .
the sdft results are compared against the ones by the exact diagonalization method @xcite . in fig .
[ fig : four ] we plot the density profiles for the system with @xmath38 and @xmath39 for @xmath77 , and @xmath78 .
we find that for @xmath79 , both the total ground - state densities @xmath62 and the spin - resolved densities show very good agreement between the sdft results and the exact diagonalization ones . when @xmath80 , sdft still keeps the overall shape of the total ground - state densities well .
however , the respective amplitudes of the spin - up and spin - down density are exaggerated by the sdft . even for this weakness ,
the sdft can help us qualitatively understand the @xmath11 crossover . for weak interaction strength ,
the interaction term gives smaller contribution comparing to the kinetic term , such that @xmath62 is dominated by the lowest occupied single - particle states .
for the system we studied here , one flat peak is found ( see fig . [
fig : four](a ) and ( b ) ) . for strong interaction strength @xmath81
@xcite , a structure consisting of @xmath82 peaks emerges ( see fig . [
fig : four](c ) and ( d ) ) , that is , @xmath62 is signaled in finite region by @xmath82 maxima . for even stronger interaction strength @xmath83
, the total atom distribution can be described by spinless fermions and the spins tend to decouple with exponentially small exchange interactions .
the system degenerates with ferromagnetic and anti - ferromagnetic states @xcite . in figs .
[ fig : five ] and [ fig : six ] , the 3d plots of the ground - state density distributions are shown as a function of the dimensionless interaction strength @xmath33 and @xmath56 . the transition from @xmath9-friedel oscillations to @xmath10-wigner oscillations can be judged by the change in the peaks in the total density distributions , which vary from @xmath84 to @xmath75 in fig .
[ fig : five ] happened at @xmath81 , and from @xmath75 to @xmath85 in fig .
[ fig : six ] happened at @xmath86 . at strongly repulsive limit
, the polarized system shows the same properties as those of the unpolarized one , where the atoms become impenetrable and behave as they were noninteracting and spinless .
dft provides a method through which we can access larger systems beyond the reach of qmc and dmrg simulations . in the following we study the polarized system confined in a trap with particle number of @xmath87 , which is not accessible by means of the present exact diagonalization method .
contour plots are shown in figs . [
fig : eight ] and [ fig : nine ] , respectively , for @xmath88 and @xmath89 . a clear signature of @xmath11 crossover is signaled by the emergence of @xmath3 peaks at the strong interaction strength . from the above discussion , we find that , in spite of the short - range nature of the interaction , density profile shows pronounced well - separated @xmath3 peaks at strong repulsive interactions .
however , the densities between the peaks do not drop to zero , which is different from the 1d system of long - range coulomb interaction , where they drop almost to zero in some sense resembling a charge density wave and a quasi - wigner molecule .
in this work , we study a 1d system of two - component fermi gases with a contact repulsive interaction .
the unpolarized or polarized atoms are either confined in a 1d box or subjected to a strongly anisotropic harmonic potential .
the total and spin - resolved densities are calculated within the bethe - ansatz based spin - density functional theory .
the results are compared to those from the exact bethe - ansatz method and the exact diagonalization scheme .
we found that the bethe - ansatz based sdft gives a faithful description for the ground state properties at weak interaction strength but exaggerates the amplitude of the density distributions especially for the corresponding spin - up and spin - down densities at strong interaction strength . in spite of the short - range nature of the interaction , a pronounced @xmath3-peak structure
is observed in the density profile in the strongly interacting regime .
for unpolarized systems , the density is characterized by @xmath0-friedel oscillations with @xmath64 peaks at the weak interaction regime . as the interaction strength increases , the number of peaks in the density profile is doubled .
this leads to the emergence of @xmath1-wigner oscillations of @xmath3 peaks in the density profile .
numerically we found a threshold interaction strength @xmath90 at which the @xmath8 transformation happens .
for polarized systems , at weak interaction strength , the peaks in the density profile depend on the system parameters .
however , when increasing the interaction strength , a threshold value can be found where @xmath3-peak oscillations appear in the density profile signaling the @xmath1-wigner oscillations . at strongly repulsive limit , the total energy and the total density profile of the system own the same properties , as those of the fully unpolarized one . there , the atoms become impenetrable and behave as they were noninteracting and spinless .
_ acknowledgments _ this work was supported by nsf of china under grant nos .
11174253 and 10974181 , zhejiang provincial natural science foundation under grant no r6110175 .
it is a pleasure to thank giovanni vignale , marco polini , shu chen , and wei li for several useful discussions .
99 t. giamarchi , _ quantum physics in one dimension _
( clarendon press , oxford , 2004 ) . h.
j. schulz , phys .
lett . * 64 * , 2831 ( 1990 ) ; * 71 * , 1864 ( 1993 ) .
e. wigner , phys . rev . * 46 * , 1002 ( 1934 ) .
erich j. mueller , phys .
b * 72 * , 075322 ( 2005 ) ; g. a. fiete , j. qian , y. tserkovnyak , and b. i. halperin , phys .
b * 72 * , 045315 ( 2005 ) .
l. shulenburger , m. casula , g. senatore , and r. m. martin , phys .
b * 78 * , 165303 ( 2008 ) .
wang , w. li , s. chen , g. xianlong , m. rontani , and m. polini , arxiv:1202.3284 , accepted to publish in phys .
astrakharchik and yu .
e. lozovik , phys .
a * 77 * , 013404 ( 2008 ) ; g.e .
astrakharchik and m.d .
girardeau , phys .
b * 83 * , 153303 ( 2011 ) .
m. machida , s. yamada , y. ohashi , and h. matsumoto , phys .
a * 74 * , 053621 ( 2006 ) ; m. machida , s. yamada , m. okumura , y. ohashi , and h. matsumoto , phys .
a * 77 * , 053614 ( 2008 ) .
r. a. molina , j. dukelsky , and p. schmitteckert , phys .
lett . * 99 * , 080404 ( 2007 ) .
m. rigol , a. muramatsu , g. g. batrouni , and r. t. scalettar , phys .
lett . * 91 * , 130403 ( 2003 ) ; m. rigol and a. muramatsu , phys . rev . a * 69 * , 053612 ( 2004 ) ;
. commun . * 243 * , 33 ( 2004 ) .
g. e. astrakharchik , d. blume , s. giorgini , and l. p. pitaevskii , phys .
lett . * 93 * , 050402 ( 2004 ) .
m. casula , d. m. ceperley , and e. j. mueller , phys .
a * 78 * , 033607 ( 2008 ) ; n. helbig , j.i .
fuks , m. casula , m.j .
verstraete , m.a.l .
marques , i.v .
tokatly , and a. rubio , phys .
a * 83 * , 032503 ( 2011 ) .
t. husslein , w. fettes , and i. morgenstern , int .
c * 8 * , 397 ( 1997 ) .
nikkarila , m. koskinen , s. m. reimann , and m. manninen , new j. phys . * 10 * , 063013 ( 2008 ) ; j .-
nikkarila , m. koskinen , and m. manninen , eur .
j. b * 64 * , 95 ( 2008 ) .
magyar and k. burke , phys .
a * 70 * , 032508 ( 2004 ) .
g. xianlong , m. polini , r. asgari , and m.p .
tosi , , 033609 ( 2006 ) .
abedinpour , m. polini , g. xianlong , and m.p .
tosi , phys .
a * 75 * , 015602 ( 2007 ) .
lima , m.f .
silva , l.n .
oliveira , and k. capelle , phys .
* 90 * , 146402 ( 2003 ) ; k. capelle , n.a .
lima , m.f .
silva , and l.n .
oliveira , in the fundamentals of electron density , density matrix and density functional theory in atoms , molecules and solids , kluwer series , `` progress in theoretical chemistry and physics , '' edited by n.i .
gidopoulos and s. wilson ( kluwer , dordrecht , 2003 ) .
g. xianlong and r. asgari , , 033604 ( 2008 ) .
d. vieira , h.j.p .
freire , v.l .
campo jr . , and k. capelle , j. mag .
mag . mat . * 320 * , e418 ( 2008 ) .
s. a. sffing , m. bortz , i. schneider , a. struck , m. fleischhauer , and s. eggert , phys .
b * 79 * , 195114 ( 2009 ) . s. a. sffing , m. bortz , and s. eggert , 021602(r ) ( 2011 ) .
g. bedrftig , b. brendel , h. frahm , and r. m. noack , phys . rev .
b * 58 * , 10225 ( 1998 ) .
b . wei , j .-
gu , h .- q .
lin , arxiv:0807.2154v1 .
l. guan , s. chen , y. wang , and z .- q .
ma , phys .
* 102 * , 160402 ( 2009 ) .
m. olshanii , , 938 ( 1998 ) ; t. bergeman , m.g .
moore , and m. olshanii , _ ibid .
_ * 91 * , 163201 ( 2003 ) .
m. gaudin , phys .
24a * , 55 ( 1967 ) ; c.n .
yang , , 1312 ( 1967 ) .
g.f . giuliani and g. vignale ,
_ quantum theory of the electron liquid _ ( cambridge university press , cambridge , 2005 )
. o. gunnarsson and b.i .
lundqvist , phys . rev .
b * 13 * , 4274 ( 1976 ) ; j.i .
fuks , a. rubio , and n.t .
maitra , phys .
a * 83 * , 042501 ( 2011 ) .
l. tonks , phys .
rev . * 50 * , 955 ( 1936 ) ; m. girardeau , j. math .
* 1 * , 516 ( 1960 ) ; b. paredes , a. widera , v. murg , o. mandel , s. flling , i. cirac , g.v .
shlyapnikov , t. w. hnsch , and i. bloch , nature * 429 * , 277 ( 2004 ) ; t. kinoshita , t. wenger , and d.s .
weiss , science * 305 * , 1125 ( 2004 ) .
k. jauregwi , w. husler , and b. kramer , europhys . lett . * 24 * 581 ( 1993 ) .
b. szafran , f.m .
peeters , s. bednarek , t. chwiej , and j. adamowski , phys .
b * 70 * , 035401 ( 2004 ) ; a. secchi and m. rontani , phys .
b * 80 * , 041404 ( 2009 ) . for the polarized system confined in a harmonic trap with @xmath91 and @xmath92 ,
the sdft based on balsda gives the threshold interaction strength at @xmath93 where the crossover from @xmath0-friedel to @xmath1-wigner oscillations happens , while the exact diagonalization gives @xmath94 .
for the unpolarized system confined in a harmonic trap with @xmath95 , the sdft based on balsda gives the threshold interaction strength at @xmath94 , while the exact diagonalization gives @xmath96 . | density oscillations of confined one - dimensional fermi gases of contact repulsive interactions in a continuous space are discussed within bethe - ansatz based spin - density functional theory .
the results are compared against the exact analytical and the exact diagonalization method . for unpolarized system ,
the number of the peaks in the density profiles is doubled signaling the crossover of the @xmath0-friedel to @xmath1-wigner oscillations ( with @xmath2 the fermi wave vector ) . for both unpolarized and polarized systems , a threshold of the short - range interaction strength
can be found where @xmath3-peak wigner oscillations of @xmath1 wave vector appear in the density profile ( @xmath3 is the total particle number ) . |
since the copula of two continuous random variables is scale - invariant , copulas are regarded as the functions that capture dependence structure between random variables . for many purposes , independence and monotone dependence
have so far been considered two opposite extremes of dependence structure . however , monotone dependence is just a special kind of dependence between two random variables .
more general complete dependence happens when functional relationship between continuous random variables are piecewise monotonic , which corresponds to their copula being a shuffle of min .
see @xcite .
mikusinski et al .
@xcite showed that shuffles of min is dense in the class of all copulas with respect to the uniform norm .
this surprising fact urged the discovery of the ( modified ) sobolev norm by siburg and stoimenov @xcite which is based on the @xmath0-operation introduced by darsow et al .
they @xcite showed that continuous random variables @xmath1 and @xmath2 are mutually completely dependent , i.e. their functional relationship is any borel measurable bijection , if and only if their copula has unit sobolev - norm .
darsow et al .
@xcite showed that for a real stochastic processes @xmath3 , the validity of the chapman - kolmogorov equations is equivalent to the validity of the equations @xmath4 for all @xmath5 , where @xmath6 denotes the copula of @xmath7 and @xmath8 .
it is then natural to investigate how dependence levels of @xmath9 and @xmath10 are related to that of @xmath11 . aside from @xmath12 , @xmath13 and @xmath14 ,
the easiest case is when @xmath9 and @xmath10 are mutual complete dependence copulas . in light of our result on denseness of shuffles of min in the mcd copulas
, we shall show that if @xmath15 then @xmath16 .
now , if @xmath17 and @xmath18 is a copula then we prove that @xmath19 coincides with a generalized shuffle of @xmath18 in the sense of durante et al .
we also give similar characterizations of shuffles of @xmath18 and generalized shuffles of min .
these characterizations have advantages of simplicity in calculations because it avoids using induced measures .
then we use this relationship to obtain a simple proof of a characterization of copulas whose orbit is singleton ( theorem 10 in @xcite ) .
note that there are many examples where shuffles of @xmath18 , i.e. @xmath19 or @xmath20 , do not have the same sobolev norm as @xmath18 .
however , we show that multiplication by unit norm copulas preserves independence , complete dependence and mutual complete dependence . since left- and right - multiplying a copula @xmath21 by unit norm copulas amount to `` shuffling '' or `` permuting '' @xmath1 and @xmath2 respectively , we introduce a new norm , called the @xmath0-norm , which is invariant under multiplication by a unit norm copula .
mutual complete dependence copulas still has @xmath0-norm one .
this invariant property implies that complete dependence copulas also possess unit @xmath0-norm .
based on the @xmath0-norm , a new measure of dependence is defined in the same spirit as the definition by siburg et al .
it turns out that this new measure of dependence satisfies most of the seven postulates proposed by rnyi @xcite .
the only known measure of dependence that satisfies all rnyi s postulates is the maximal correlation coefficient .
this manuscript is structured as follows .
we shall summarize related basic properties of copulas , the binary operator @xmath0 and the sobolev norm in section 2 .
then we obtain a characterization of copulas with unit sobolev norm which implies that the @xmath0-product of mcd copulas is a mcd copula in section 3 .
section 4 contains our characterizations of generalized shuffles of min and ( generalized ) shuffles of copulas in the sense of durante et al . in terms of the @xmath0-product .
we then show that shuffling a copula preserves independence , complete dependence and mutual complete dependence . in section 5 ,
a new norm is introduced and its properties are proved . and in section 6 , we define a new measure of dependence and verify that it satisfies most of rnyi s postulates .
a _ bivariate copula _ is defined to be a joint distribution function of two random variables with uniform distribution on @xmath22.$ ] since such a joint distribution is uniquely determined by its restriction on @xmath22 ^ 2 $ ] one can also define a copula as a function @xmath23 ^ 2 \to [ 0 , 1]$ ] satisfying the following properties @xmath24 @xmath25 for all @xmath26 ^ 2 $ ] and @xmath27 ^ 2 $ ] such that @xmath28 note that the two definitions are equivalent .
every copula @xmath18 induces a measure @xmath29 on @xmath30 ^ 2 $ ] by @xmath31\times[y , v ] ) = c(u , v)-c(u , y)-c(x , v)+c(x , y).\ ] ] the induced measure @xmath29 is doubly stochastic in the sense that for every borel set @xmath10 , @xmath32\times b ) = m(b ) = \mu_c(b\times[0,1])$ ] where @xmath33 is lebesgue measure on @xmath34 .
important copulas include the frchet - hoeffding upper and lower bounds @xmath35 and the product , or independent , copula @xmath36 .
a fundamental property is that @xmath13 is a copula of @xmath1 and @xmath2 if and only if @xmath2 is almost surely an increasing bijective function of @xmath1 .
if @xmath1 and @xmath2 are uniformly distributed on @xmath30 $ ] then that bijection is the identity map on @xmath30 $ ] .
its graph , the main diagonal , is the support of the induced measure @xmath37 , also called the support of @xmath13 . at the other extreme
, the minimum copula @xmath38 corresponds to random variables being monotone decreasing function of each other .
listed below are some basic properties of any copula @xmath18 , some of which shall be used frequently in the manuscript . 1
. @xmath39 for all @xmath40.$ ] 2 .
@xmath41 @xmath42 ^ 2 $ ] and hence @xmath18 is uniformly continuous .
3 . @xmath43 and @xmath44 exist almost everywhere on @xmath30 ^ 2 $ ] .
4 . for a.e . @xmath45 $ ] , @xmath46 is nondecreasing in the domain where it exists and similar statement holds for @xmath47 .
perhaps , the most important property of copulas is given by the sklar s theorem which states that to every joint distribution function @xmath48 of continuous random variables @xmath1 and @xmath2 with marginal distributions @xmath49 and @xmath50 respectively , there corresponds a unique copula @xmath18 , called the _ copula of @xmath1 and @xmath2 _ for which @xmath51 for all @xmath52 this means that the copula of @xmath53 captures all dependence structure of the two random variables . @xmath1 and @xmath2 are said to be _ mutually completely dependent _ if there exists an invertible borel measurable function @xmath54 such that @xmath55 shuffles of min were introduced by mikusinski et al .
@xcite as examples of copulas of mutually completely dependent random variables . by definition ,
a shuffle of min is constructed by shuffling ( permuting ) the support of the min copula @xmath13 on @xmath56 vertical strips subdivided by a partition @xmath57 .
it is shown ( * ? ? ?
* theorems 2.1 & 2.2 ) that the copula of @xmath1 and @xmath2 is a shuffle of min if and only if there exists an invertible borel measurable function @xmath54 with finitely many discontinuity points such that @xmath58 . in @xcite , such an @xmath54
is called strongly piecewise monotone function .
following @xcite , the binary operation @xmath0 on the set @xmath59 of all bivariate copulas is defined as @xmath60\ ] ] and the _ sobolev norm _ of a copula @xmath18 is defined by @xmath61 it is well - known that @xmath62 is a monoid with null element @xmath12 and identity @xmath13 .
so a copula @xmath18 is called _ left invertible _
( _ right invertible _ ) if there is a copula @xmath63 for which @xmath64 ( @xmath65 ) .
it was shown in ( * ? ? ?
* theorem 7.6 ) and ( * ? ? ?
* theorem 4.2 ) that the @xmath0-product on @xmath59 is jointly continuous with respect to the sobolev norm but not with respect to the uniform norm .
moreover , they @xcite gave a statistical interpretation of the sobolev norm of a copula .
[ thm : normc ] let @xmath18 be a bivariate copula of continuous random variables @xmath1 and @xmath2 .
@xmath66 ; 2 . )
@xmath67 if and only if @xmath68 ; and 3 . )
the following are equivalent .
1 . @xmath69 .
2 . @xmath18 is invertible with respect to @xmath0 .
[ thm : unitnorm3 ] for each @xmath70 $ ] , @xmath71 a.e .
4 . there exists a borel measurable bijection @xmath72 such that @xmath73 a.e .
it follows readily that all shuffles of min have norm one .
let @xmath18 be a copula with unit sobolev norm .
then @xmath74 and @xmath75 take values @xmath76 or @xmath77 almost everywhere .
see , for example , theorem 7.1 in @xcite and theorem 4.2 in @xcite .
let us recall from ( * ? ? ?
* theorem 2.2.7 ) that , for a.e .
@xmath78 , \partial_1 c(x , y)$ ] is a nondecreasing function of @xmath79.$ ] similar statement holds also for @xmath80 so for a.e .
@xmath78,$ ] there is @xmath81 $ ] such that for almost every @xmath82 , @xmath83 if @xmath84 and @xmath85 if @xmath86 ( @xmath87 ) denote the set of such @xmath88 s by @xmath89 so that @xmath90 . and for every @xmath91 , by redefining @xmath92 on a set of measure zero , we may assume that @xmath92 is defined and nondecreasing for all @xmath93 $ ] . to show that @xmath54 is measurable ,
let @xmath94,$ ] and observe that since @xmath92 is increasing in @xmath82 @xmath95 which is measurable because each @xmath96 is measurable . in exactly the same fashion
, there exists a measurable function @xmath97 \to [ 0 , 1]$ ] for which @xmath98 let us recall the definition that a measurable function @xmath99\to[0,1]$ ] is said to be _ measure - preserving _
if @xmath100 for any lebesgue measurable set @xmath101 $ ] .
[ thm : suppcopulanormone ] suppose @xmath18 is a copula with unit sobolev norm
. then there exists a unique invertible borel measurable function @xmath102 \to [ 0 , 1]$ ] such that @xmath54 is measure - preserving and for almost every @xmath103 in @xmath30 ^ 2 $ ] @xmath104 furthermore , if @xmath54 is continuous on an interval @xmath105 then it is differentiable on @xmath105 with constant derivative equal to either @xmath77 or @xmath106 . during the preparation of this manuscript , we have come across similar results such as proposition 1 in @xcite and
theorem 2.4 and corollary 2.4.1 in @xcite .
we first claim that @xmath54 and @xmath107 defined above are inverses of each other in the sense that @xmath108 and @xmath109 are identity on @xmath22 $ ] a.e . ,
i.e. @xmath110 and @xmath111 both have measure @xmath77 .
this is equivalent to saying that @xmath112 if and only if @xmath113 for a.e . @xmath114 ^ 2 $ ]
. indeed , observe that for any open interval @xmath115 $ ] , @xmath116 if and only if @xmath117 .
now let @xmath118 and @xmath119 be open intervals in @xmath30 $ ] for which @xmath120 does not intersect the graph @xmath112 , i.e. @xmath121 , hence @xmath92 is independent of @xmath122 for a.e .
so @xmath124 for a.e .
@xmath123 and all @xmath122 .
then for @xmath125 in @xmath126 @xmath127 and so @xmath128 is independent of @xmath123 which implies that @xmath120 does not intersect the graph @xmath113 .
the converse can be shown by a similar argument .
since the graph of a borel function is a borel subset of @xmath30 ^ 2 $ ] , @xmath112 and @xmath113 give the same graph .
and the claim follows .
let @xmath29 denote the doubly stochastic measure associated with @xmath18 .
a straightforward verification gives @xmath129 for all borel rectangles @xmath120 , which implies by a standard measure - theoretic technique that holds for all borel sets @xmath130 $ ] .
so @xmath54 is measure - preserving since it is equivalent to the validity of for all borel sets @xmath9 and @xmath10 .
lastly , we prove that if @xmath54 is continuous on an open interval @xmath131 then it is differentiable with @xmath132 being constant and equal to @xmath133 .
since @xmath54 is continuous and one - to - one on @xmath105 , it has to be strictly monotonic on @xmath105 .
let us consider the case where @xmath54 is strictly increasing on @xmath105 .
this implies that @xmath134 is strictly increasing on the interval @xmath135 and that @xmath84 if and only if @xmath136 for @xmath137 . for @xmath138\subset ( a , b),$ ] @xmath139 since @xmath140 and @xmath141 are differentiable with respect to @xmath82 almost everywhere , we have for a.e
. @xmath82 , @xmath142 as @xmath143 and @xmath75 and @xmath144 are equal to @xmath76 or @xmath77 , @xmath145 and hence @xmath146 for all @xmath137 .
similarly , if @xmath54 is strictly decreasing on @xmath147 then @xmath148 a.e . on @xmath147
a natural question is then to investigate the set on which an invertible measure - preserving function @xmath54 is continuous .
unfortunately , the support of a unit norm copula may be the graph of a function which is discontinuous on a dense subset of @xmath30 $ ] , and hence there is no interval on which it is continuous .
define a sequence of shuffles of min @xmath149 by letting @xmath150 be the comonotonic copula supported on the main diagonal .
@xmath151 is defined so that it shares the same support with @xmath150 in @xmath152\times[0,1]$ ] and its support in the other half @xmath153=[\frac{1}{2},1]\times[0,1]$ ] is that of @xmath150 flipped horizontally .
@xmath154 is then obtained from @xmath151 by flipping the support in each stripe of the set @xmath155 $ ] where @xmath156\cup{\left([\frac{1}{2 ^ 2},\frac{1}{2}]+\frac{1}{2}\right)}$ ] .
for general @xmath157 , we define @xmath158 and let the shuffle of min @xmath159 be obtained from @xmath160 by flipping the support in each stripe of @xmath161 horizontally . to sum up , each shuffle of min @xmath159 is supported on the graph @xmath162 where @xmath163 is constructed according to the above iterative procedure , starting from @xmath164 and @xmath165}$ ] . the first few @xmath159 s are illustrated in figure [ fig : s_n ] .
[ fig : s_n ] @[email protected]@ , @xmath154 , and @xmath166,title="fig:",scaledwidth=26.0% ] , @xmath154 , and @xmath166,title="fig:",scaledwidth=26.0% ] & , @xmath154 , and @xmath166,title="fig:",scaledwidth=80.0% ] from construction , @xmath161 consists of @xmath167 stripes , each of width @xmath168 . on each of these stripes ,
the supports of @xmath159 and @xmath160 differ by a flip which implies that @xmath169 and @xmath170 are equal on the stripe except on two triangles of total area @xmath171 where @xmath172 . similarly , on each stripe of @xmath161 , @xmath173 on two triangles of total area @xmath174 and zero elsewhere .
therefore , @xmath175 now , given @xmath176 , @xmath177 which converges to @xmath76 as @xmath178 . since the set of copulas is complete with respect to the sobolev norm ( see p. 424
in @xcite ) , the cauchy sequence @xmath149 converges to a copula @xmath179 .
it follows that @xmath180 .
it can also be shown that the support of @xmath179 contains the graph of the pointwise limit @xmath54 of @xmath163 .
finally , we shall show that the mutual complete dependence copula @xmath179 has support on the graph of a function discontinuous on the set of dyadic points in @xmath30 $ ] .
in fact , it is straightforward to calculate the jump of @xmath54 at a dyadic point @xmath181 where @xmath182 is indivisible by @xmath183 : @xmath184 we note here that the support of @xmath179 is self - similar with hausdorff dimension one . a surprising fact by mikusinski , sherwood and taylor ( * ? ? ?
* theorem 3.1 ) is that every copula , in particular the independence copula , can be approximated arbitrarily close in the uniform norm by a shuffle of min .
consequently , the uniform norm can not distinguish dependence structures among copulas .
however , if @xmath149 is a sequence of shuffles of min converging in the sobolev norm to a copula @xmath18 , then it is necessary that @xmath69 , hence @xmath18 is a copula of two mutually completely dependent random variables .
conversely , one might ask whether any copula @xmath18 with @xmath69 can be approximated arbitrarily close in the sobolev norm by a shuffle of min .
we quote here without proof a result from @xcite which will be useful in answering the question .
[ thm : chounguyen ] for every measure - preserving function @xmath54 over @xmath30 $ ] , there exists a sequence of bijective piecewise linear measure - preserving functions @xmath185 whose slopes are either @xmath186 or @xmath106 and such that @xmath163 converges to @xmath54 a.e .
[ lem : norm - supp ] let @xmath187 and @xmath188 be copulas with norm one which are supported on the graphs of @xmath189 and @xmath190 , respectively
. then @xmath191 by assumption , for a.e .
@xmath103 , @xmath192 if and only if @xmath82 is between @xmath193 and @xmath194 . likewise , @xmath195 if and only if @xmath88 is between @xmath196 and @xmath197 for a.e .
so @xmath198 [ thm : c - s ] for any copula @xmath18 with @xmath199 , there exists a sequence of shuffles of min @xmath149 such that @xmath200 .
suppose @xmath18 is a copula with norm one and @xmath18 is supported on the graph of @xmath54 .
it follows from theorem [ thm : suppcopulanormone ] that @xmath54 is a measure - preserving bijection from @xmath30 $ ] onto itself . by theorem [ thm : chounguyen ]
, one can construct a sequence of measure - preserving functions @xmath185 for which each @xmath163 is bijective piecewise linear with slopes @xmath186 or @xmath106 and @xmath163 converges to @xmath54 a.e . a corresponding sequence of shuffles of min @xmath201
can then be chosen so that the graph of @xmath163 is the support of @xmath159 . by lemma [ lem : norm - supp ] , @xmath202 . since @xmath203 a.e .
, an application of dominated convergence theorem shows that @xmath204 .
consequently , @xmath205 in the sobolev norm . from the proof ,
it is worth noting that one can approximate a copula @xmath18 by only straight shuffles of min whose slopes on all subintervals are @xmath186 .
let @xmath206 . 1 .
if @xmath207 and @xmath208 then @xmath209 .
2 . if @xmath210 then @xmath207 if and only if @xmath208 .
[ lem1 ] \1 .
let @xmath206 be such that @xmath207 and @xmath208 . by theorem [ thm : c - s ]
, there exist sequences @xmath149 , @xmath211 of shuffles of min such that @xmath212 and @xmath213 in the sobolev norm . hence , with respect to the sobolev norm ,
@xmath214 by the joint continuity of the @xmath0-product .
since a product of shuffles of min is still a shuffle of min , @xmath215 \2 . let @xmath216 and @xmath217 be copulas of sobolev norm 1 . since @xmath218 , an application of 1 .
at least as soon as shuffles of min were introduced in @xcite , the idea of simple shuffles of copulas was already apparent .
see , e.g. , @xcite . in @xcite , durante , sarkoci and
sempi gave a general definition of shuffles of copulas via a characterization of shuffles of min in terms of a _ shuffling _
@xmath220 ^ 2\to [ 0,1]^2 $ ] defined by @xmath221 where @xmath222\to [ 0,1]$ ] . before stating their results ,
let us recall the definition of push - forward measures .
let @xmath54 be a measurable function from a measure space @xmath223 to a measurable space @xmath224 .
push - forward of @xmath225 under @xmath54 _ is the measure @xmath226 on @xmath224 defined by @xmath227 for @xmath228 .
a copula @xmath18 is a shuffle of min if and only if there exists a piecewise - continuous measure - preserving bijection @xmath229\to[0,1]$ ] such that @xmath230 . dropping piecewise continuity of @xmath231 , a
_ generalized shuffle of min _ is defined as a copula @xmath18 whose induced measure is @xmath230 for some measure - preserving bijection @xmath229\to[0,1]$ ] . replacing @xmath13 by a given copula @xmath63 , a _ shuffle of @xmath63 _ is a copula @xmath18 whose induced measure is @xmath232 for some piecewise - continuous measure - preserving bijection @xmath231 .
@xmath18 is also called the _ @xmath231-shuffle of @xmath63_. if the bijection @xmath231 is only required to be measure - preserving in , then @xmath18 is called a _ generalized shuffle of @xmath63_. the following lemma will be useful in our investigation .
[ lem:1 ] let @xmath231 be a measure - preserving bijection on @xmath30 $ ] and @xmath18 be a copula defined by @xmath233\times[0,y]\right)}\quad\text{for } x , y\in[0,1].\ ] ] then the copula @xmath18 , or equivalently its induced measure @xmath234 , is supported on the graph of @xmath235 .
moreover , the converse also holds , i.e. if @xmath18 is supported on the graph of a measure - preserving bijection @xmath231 then @xmath236 .
let @xmath237\times[c , d]$ ] be a closed rectangle in @xmath238 and @xmath239 be the map on @xmath30 ^ 2 $ ] associated with a given measure - preserving bijection @xmath231 on @xmath30 $ ] , i.e. @xmath240 .
so @xmath241\times[c , d]\right ) } = { \left(t^{-1}[a , b]\right)}\times[c , d]$ ] and , by definition of the push - forward measure , @xmath242\times[c , d]\right ) } & = \mu_m{\left(s_t^{-1}{\left([a , b]\times[c , d]\right)}\right)}\\ & = \mu_m{\left({\left(t^{-1}[a , b]\right)}\times[c , d]\right ) } = m{\left({\left(t^{-1}[a , b]\right)}\cap[c , d]\right)}. \end{aligned}\ ] ] thus , @xmath243\times[c , d]\right ) } = 0 $ ] if and only if the projection of @xmath244\times[c , d]\right)}$ ] onto @xmath245 $ ] has measure zero .
consequently , since borel measurable subsets of @xmath30 ^ 2 $ ] are generated by rectangles , the desired result is obtained .
a copula @xmath18 is a generalized shuffle of min if and only if @xmath69 . ( @xmath246 ) let @xmath18 be a generalized shuffle of min , i.e. there exists a measure preserving bijection @xmath231 on @xmath30 $ ] such that @xmath247 . by theorem [ thm : chounguyen ] , there is a sequence @xmath211 of piecewise - continuous measure - preserving bijection on @xmath30 $ ] such that @xmath248 a.e .
so @xmath249\times[0,y]\right)}$ ] defines a sequence of shuffles of min .
we claim that @xmath250 .
in fact , by lemma [ lem:1 ] , @xmath251 and @xmath252 are supported on the graphs of @xmath235 and @xmath253 respectively .
now , lemma [ lem : norm - supp ] implies that @xmath254 which converges to @xmath76 as a result of the lusin - souslin theorem ( see , e.g. , ( * ? ? ?
* corollary 15.2 ) ) which states that a borel measurable injective image of a borel set is a borel set and the dominated convergence theorem .
therefore , @xmath255 in the sobolev norm .
( @xmath256 ) let @xmath18 be a copula with @xmath69 .
then theorem [ thm : suppcopulanormone ] gives a measure - preserving bijection @xmath54 whose graph is the support of @xmath18 .
so lemma [ lem:1 ] implies that @xmath257 .
[ thm : mu*nu ] if @xmath225 and @xmath258 are doubly stochastic measures on @xmath30 ^ 2 $ ] then @xmath259 induces a doubly stochastic ( borel ) measure @xmath260 on @xmath30 ^ 2 $ ] , where @xmath261)\quad\text{and}\quad \partial_1\nu(t , j ) = \frac{d}{dt}\nu([0,t]\times j).\ ] ] furthermore , if @xmath9 and @xmath10 are copulas and @xmath262 and @xmath263 denote their doubly stochastic measures then @xmath264 we shall prove only which shows that @xmath260 is a doubly stochastic measure when the measures @xmath225 and @xmath258 are doubly stochastic and inducible by copulas .
let @xmath9 and @xmath10 be copulas and @xmath265 , j=[b_1,b_2 ] \subseteq [ 0,1]$ ]
. then @xmath266 \,dt\\ & = \int_0 ^ 1 \partial_2{\left(a(a_2,t)-a(a_1,t)\right)}\partial_1{\left(b(t , b_2)-b(t , b_1\right)}\,dt\\ & = \int_0 ^ 1 \frac{d}{dt}\mu_a(i\times[0,t])\frac{d}{dt}\mu_b([0,t]\times j)\,dt\\ & = \mu_a*\mu_b(i\times j ) .
\end{aligned}\ ] ] the usual measure - theoretic techniques allow to extend this result to the product of all borel sets .
[ lem : s_tassoc ] let @xmath231 be a measure - preserving bijection on @xmath30 $ ] and @xmath225 , @xmath258 be doubly stochastic measures on @xmath30 ^ 2 $ ] .
then @xmath267 let @xmath105 and @xmath268 be borel sets in @xmath30 $ ]
. then @xmath269 [ thm : socchar ] let @xmath18 and @xmath63 be bivariate copulas
. then 1 .
@xmath18 is a shuffle of @xmath63 if and only if there exists a shuffle of min @xmath9 such that @xmath270 ; 2 .
@xmath18 is a generalized shuffle of @xmath63 if and only if there exists a generalized shuffle of min @xmath9 such that @xmath270 .
we shall only prove 2 .
since 1 .
is just a special case .
( @xmath246 ) if @xmath18 is a shuffle of @xmath63 , i.e. @xmath271 for some measure - preserving bijection @xmath231 of @xmath30 $ ] , then the copula @xmath9 defined by @xmath272 is a shuffle of min by theorem [ thm : mu*nu ] .
then @xmath273 which means that @xmath274 .
( @xmath256 ) if @xmath274 for some copula @xmath9 with @xmath17 then @xmath272 for some measure - preserving bijection @xmath231 and @xmath275 note the repeated uses of theorem [ thm : mu*nu ] and lemma [ lem : s_tassoc ] in both derivations .
since @xmath12 is the only null element of @xmath0 ( see @xcite ) , it follows easily from theorem [ thm : socchar ] that @xmath12 is the only copula which is invariant under shuffling by generalized shuffles of min .
this is a result first proved in ( * ? ? ?
* theorem 10 ) . even though all generalized shuffles of min have equal unit norm , not all shuffles of @xmath18 have the same norm . here
is a class of examples .
[ exam : shufflediffnorm ] for @xmath276 , let @xmath277 denote the straight shuffle of min whose support is on the main diagonals of the squares @xmath278\times[1-\alpha,1]$ ] and @xmath279\times[0,1-\alpha]$ ] .
then by straightforward computations , for any copula @xmath18 , @xmath280 and @xmath281 let us now consider the farlie - gumbel - morgenstern ( fgm ) copulas @xmath282 , @xmath283 $ ] , defined by @xmath284
. then @xmath285 and @xmath286 so that @xmath287 which is equal to @xmath288 only if @xmath289 or @xmath290 or @xmath77 .
for each @xmath291 , @xmath292 is maximized when @xmath293 and the maximum value is @xmath294 .
if @xmath295 and @xmath2 are conditionally independent given @xmath1 , then @xmath296 [ da - con - ind ] let @xmath297 be borel measurable and @xmath298 be random variables
. then @xmath299 and @xmath2 are conditionally independent given @xmath1.[special - case ] since @xmath72 is borel measurable , @xmath299 is measurable with respect to @xmath300 , the @xmath301-algebra generated by @xmath1 .
hence , by properties of conditional expectations , @xmath302 for all @xmath303 .
this completes the proof .
[ cor : star - shuffling ] let @xmath304 be borel measurable functions . then @xmath305 for all random variables @xmath298.[decompose ]
since @xmath54 and @xmath107 are borel measurable , by propositions [ da - con - ind ] and [ special - case ] , we have @xmath306 for all random variables @xmath298 .
transposing both sides of , we obtain @xmath307 then , we have @xmath308 let @xmath309 , the set of invertible copulas or , equivalently , the set of copulas with unit sobolev norm .
a _ shuffling map _
@xmath310 is a map on @xmath311 defined by @xmath312 .
the motivation behind the word shuffling comes from the fact that a shuffling image of a copula is a two - sided generalized shuffle of the copula .
note that @xmath313 let @xmath298 be continuous random variables and @xmath309 .
then the following statements hold : 1 . @xmath1 and @xmath2 are independent if and only if @xmath314 .
@xmath1 is completely dependent on @xmath2 or vice versa if and only if @xmath315 is a complete dependence copula .
3 . @xmath1 and @xmath2 are mutually completely dependent if and only if @xmath315 is a mutual complete dependence copula .
[ dependence - invariant ] \1 .
this clearly follows from the fact that @xmath12 is the zero element in @xmath316 .
2 . with out loss of generality ,
let us assume that @xmath2 is completely dependent on @xmath1 , i.e. there exists a borel measurable transformation @xmath72 such that @xmath73 with probability one .
let @xmath54 and @xmath107 be borel measurable bijective transformations on @xmath34 such that @xmath317 and @xmath318 . by corollary
[ decompose ] , we have @xmath319 .
thus , it suffices to show that @xmath320 is completely dependent on @xmath321 . from @xmath73 with probability one , @xmath322 with probability one .
it is left to show that @xmath323 is borel measurable .
this is true because of lusin - souslin theorem ( see , e.g. , @xcite , corollary 15.2 ) which states that a borel measurable injective image of a borel set is a borel set .
the converse automatically follows because the inverse of a shuffling map is still a shuffling map .
the proof is completely similar to above except that the function @xmath72 is also required to be bijective .
corollary [ decompose ] implies that a shuffling image of a copula @xmath324 is a copula of transformed random variables @xmath325 for some borel measurable bijective transformations @xmath54 and @xmath107 . together with the above lemma , we obtain the following theorem .
let @xmath1 and @xmath2 be continuous random variables .
let @xmath54 and @xmath107 be any borel measurable bijective transformations of the random variables @xmath1 and @xmath2 , respectively
. then @xmath1 and @xmath2 are independent , completely dependent or mutually completely dependent if and only if @xmath321 and @xmath320 are independent , completely dependent or mutually completely dependent , respectively .
the above theorem suggests that shuffling maps preserve stochastic properties of copulas . in the next section ,
we contruct a norm which , in some sense , also preserves stochastic properties of copulas .
our main purpose is to construct a norm under which shuffling maps are isometries and then derive its properties .
define a map @xmath326 , by @xmath327 by straightforward verifications , @xmath328 is a norm on @xmath311 , called the _
@xmath0-norm_. moreover , it is clear from the definition that @xmath329 for all @xmath330 .
the following proposition summarizes basic properties of the @xmath0-norm .
observe that properties 2.4 .
are the same as those for the sobolev norm .
[ prop:*norm ] let @xmath331 .
then the following statements hold .
1 . @xmath332 if @xmath333 .
2 . @xmath334 if and only if @xmath68 .
[ prop:*norm3 ] @xmath335 .
@xmath336 for all @xmath330 .
[ props ] 1 . is a consequence of the inequality @xmath337 .
2 . follows from the fact that @xmath12 is the zero of @xmath316 . to prove 3 .
, we first observe that @xmath338 for all @xmath309 .
the result follows by taking supremum over @xmath309 on both sides . finally , using the facts that @xmath339 for all @xmath340 , @xmath341 let @xmath342 and @xmath343
. then @xmath344 .
therefore , shuffling maps are isometries with respect to the @xmath0-norm .
we shall prove only one side of the equation as the other can be proved in a similar fashion .
let @xmath345 and @xmath346 .
then by corollary [ lem1 ] , for any @xmath331 , @xmath347 if and only if @xmath348 .
hence , @xmath349 from example [ exam : shufflediffnorm ] , let @xmath350 , @xmath283\setminus{\left\{0\right\}}$ ] and @xmath351 .
then @xmath352 .
since @xmath353 for any @xmath354 .
then @xmath355 hence , the sobolev norm and the @xmath356-norm are distinct . [ norms - distinct ] [ exam : equalnorms ] let @xmath357 $ ] and @xmath18 be a copula .
recall that one can show using only the property @xmath358 of the norm @xmath359 ( see @xcite ) that @xmath360 .
since the @xmath0-norm shares this same property with the sobolev norm ( see proposition [ prop:*norm]([prop:*norm3 ] ) ) , we also have @xmath361 so @xmath362 for all copulas @xmath18 satisfying @xmath363 .
in particular , the sobolev norm and the @xmath356-norm coincide on the family of convex sums of an invertible copula and the product copula , where the norms are equal to @xmath364 .
[ lem : shufflingcd ] let @xmath365 $ ] be a borel measurable set . define the function @xmath366 \to [ 0,1]$ ] by @xmath367\cap a ) & \text{if } x\in a,\\ m(a ) + m([0,x]\setminus a ) & \text{if } x\notin a. \end{cases}\ ] ] then @xmath368 is measure - preserving and _ essentially invertible _ in the sense that there exists a borel measurable function @xmath369 for which @xmath370 a.e .
@xmath45 $ ] .
such a @xmath369 is called an _
essential inverse _ of @xmath368 .
clearly , @xmath368 is borel measurable .
* @xmath371 @xmath368 is measure - preserving : * it suffices to prove that @xmath372)=m([0,b])$ ] for all @xmath373 $ ] .
now if @xmath374 , then @xmath375 = \ { x\in a\colon s(x ) \in [ 0,b ] \ } = { \left\ { x \in a\colon m(a \cap [ 0,x ] ) \leq b \right\}}.\ ] ] by continuity of @xmath33 , there exists a largest @xmath376 such that @xmath377 ) = b$ ] .
then @xmath378 = a \cap \{x\colon m(a \cap [ 0,x ] ) \leq b\ } = a \cap [ 0,x_0]$ ] .
therefore , @xmath372 ) = m(a \cap [ 0,x_0 ] ) = b$ ] .
the case where @xmath379 can be proved similarly .
* @xmath371 @xmath368 is essentially invertible : * using continuity of @xmath33 , we shall define an auxiliary function @xmath369 on @xmath30 $ ] as follows .
if @xmath380 , there exists a corresponding @xmath123 such that @xmath381\cap a\right ) } = y$ ] .
if @xmath382 , there exists a corresponding @xmath383 such that @xmath384\setminus a\right ) } = y$ ] . in these two cases ,
we define @xmath385 .
generally , @xmath369 is not unique as there might be many such @xmath88 s .
we shall show that @xmath368 is injective outside a borel set of measure zero by proving that @xmath386 is the identity map on @xmath30\setminus z$ ] for some borel set @xmath295 of measure zero . if @xmath123 then @xmath387 so that @xmath388 and @xmath389\cap a ) = s_a(x ) = m([0,x]\cap a)$ ] .
similarly , if @xmath383 then @xmath390 , @xmath391 and @xmath389\setminus a ) = m([0,x]\setminus a)$ ] . consider the set of all @xmath123 for which @xmath392 .
if @xmath393 then @xmath394\cap a\right ) } = 0 $ ] which implies that @xmath395 wherever exists .
now , a simple application of the radon - nikodym theorem ( see , e.g. , @xcite ) yields that the limit in is equal to @xmath77 for all @xmath396 where @xmath295 is a borel set of measure zero .
therefore , @xmath397 has zero measure .
similarly , @xmath398 is a null set and hence @xmath399 .
therefore , @xmath400 except possibly on a borel set @xmath295 of measure zero . by the lusin - souslin theorem ( see , e.g. , ( * ? ? ?
* corollary 15.2 ) ) , the injective borel measurable function @xmath368 , mapping a borel set @xmath30\setminus z$ ] onto a borel set @xmath401\setminus z)$ ] , has a borel measurable inverse , still denoted by @xmath369 .
now , since @xmath368 is measure - preserving , its range which is the domain of @xmath369 has full measure .
this guarantees that @xmath369 can be extended to a borel measurable function on @xmath30 $ ] which is an essential inverse of @xmath368 .
[ thm : cdnorm1 ] let @xmath298 be random variables on a common probability space for which @xmath2 is completely dependent on @xmath1 or @xmath1 is completely dependent on @xmath2
. then @xmath402 .
assume that @xmath2 is completely dependent on @xmath1 .
then @xmath21 is a complete dependence copula for which @xmath403 where @xmath216 is a uniform random variable on @xmath30 $ ] and @xmath404 \rightarrow [ 0,1]$ ] is a measure - preserving borel function .
note that @xmath405 is also a uniform random variable and that @xmath18 is left invertible .
as the first step , we shall construct an invertible copula @xmath151 such that @xmath406 is supported in the two diagonal squares @xmath407 ^ 2 \cup [ 1/2,1]^2 $ ] .
let @xmath408)$ ] , denote @xmath409 as defined in and put @xmath410 . by lemma [ lem : shufflingcd ] , @xmath411 is invertible a.e . and
hence @xmath151 is invertible . by corollary [ cor : star - shuffling ] ,
@xmath412 where @xmath413 is still a uniform random variable on @xmath30 $ ] .
it is left to verify that the support of @xmath406 lies entirely in the two diagonal squares which can be done by showing that the graph of @xmath414 is contained in the area .
in fact , since @xmath415 ) \subseteq a$ ] , it follows that @xmath416 ) \subseteq f(a ) \subseteq [ 0,\frac{1}{2}]$ ] .
the inclusion @xmath417 ) \subseteq [ \frac{1}{2},1]$ ] can be shown similarly . as a consequence
, @xmath406 can be written as an ordinal sum of two copulas , @xmath187 and @xmath188 , with respect to the partition @xmath418,[\frac{1}{2},1]\}$ ] . since left - multiplying a copula @xmath18 by an invertible copula amounts to shuffling the first coordinate of @xmath18
, it follows that @xmath187 and @xmath188 are still supported on closures of graphs of measure - preserving functions .
next , we apply the same process to @xmath187 and @xmath188 which yields invertible copulas @xmath419 and @xmath420 for which @xmath421 and @xmath422 are both supported in @xmath152 ^ 2 \cup [ \frac{1}{2},1]^2 $ ] and define @xmath154 to be the ordinal sum of @xmath419 and @xmath420 with respect to the partition @xmath418,[\frac{1}{2},1]\}$ ] .
@xmath154 is again an invertible copula .
then the support of @xmath423 is contained in the four diagonal squares @xmath424}^2 $ ] .
therefore , @xmath423 is an ordinal sum with respect to the partition @xmath425,[\frac{1}{4},\frac{1}{2}],[\frac{1}{2},\frac{3}{4}],[\frac{3}{4},1]\}$ ] of four copulas each of which is supported on the closure of graph of a measure - preserving function . by successively applying this process , we obtain a sequence of invertible copulas @xmath426 , defined by @xmath427 , such that the support of @xmath428 is a subset of the @xmath167 diagonal squares .
so @xmath429 pointwise outside the main diagonal and so are their partial derivatives .
hence @xmath430 .
thus @xmath431 .
if @xmath1 is completely dependent on @xmath2 then @xmath21 is right invertible and similar process where suitably chosen @xmath159 s are multiplied on the right yields a sequence @xmath432 of invertible copulas such that @xmath433 as desired [ cor : l*r ] let @xmath434 and @xmath435 be left invertible and right invertible copulas , respectively
. then @xmath436 . from theorem [ thm : cdnorm1 ] , there exist sequences of invertible copulas @xmath159 and @xmath437 such that @xmath438 and @xmath439 in the sobolev norm . by the joint continuity of the @xmath0-product with respect to the sobolev norm , @xmath440
. therefore , @xmath441 .
let us give some examples of copulas of the form @xmath442 .
consider a copula @xmath18 , @xmath443 and @xmath444 whose supports are shown in the figure below .
( -0.25,-0.25)(4.2,1.2 ) ( 0,1)(1,1 ) ( 1,0)(1,1 ) ( 0,0)(0,1 ) ( 0,0)(1,0 ) ( 1.5,1)(2.5,1 ) ( 2.5,0)(2.5,1 ) ( 1.5,0)(1.5,1 ) ( 1.5,0)(2.5,0 ) ( 3,1)(4,1 ) ( 4,0)(4,1 ) ( 3,0)(3,1 ) ( 3,0)(4,0 ) ( 0.5,0)(0.5,1 ) ( 1.5,0.5)(2.5,0.5 ) ( 3,0.5)(4,0.5 ) ( 3.5,0)(3.5,1 ) as mentioned before , the copula @xmath444 , though neither left nor right invertible , has unit @xmath0-norm .
in @xcite , rnyi triggered numerous interests in finding the `` right '' sets of properties that a natural ( if any ) measure of dependence @xmath445 should possess .
for reference , the seven postulates proposed by rnyi are listed below . 1 . [
renyi : def ] @xmath445 is defined for all non - constant random variables @xmath1 , @xmath2 .
[ renyi : sym ] @xmath446 .
[ renyi:01 ] @xmath447 $ ] .
[ renyi:0 ] @xmath448 if and only if @xmath1 and @xmath2 are independent .
[ renyi:1 ] @xmath449 if either @xmath450 or @xmath451 a.s . for some borel - measurable functions @xmath54 , @xmath107
[ renyi : scale ] if @xmath452 and @xmath453 are borel bijections on @xmath34 then @xmath454 .
[ renyi : rho ] if @xmath1 and @xmath2 are jointly normal with correlation coefficient @xmath455 , then @xmath456 .
recently , siburg and stoimenov @xcite introduced a measure of mutual complete dependence @xmath457 defined via its copula @xmath324 by @xmath458 .
while @xmath459 is defined only for continuous random variables , it satisfies the next three properties [ renyi : sym].[renyi:0 ] .
enjoyed by most if not all measures of dependence .
however , instead of the conditions [ renyi:1 ] . and [ renyi : scale ] .
, @xmath459 satisfies the following conditions which makes it suitable for capturing mutual complete dependence regardless of how the random variables are related . * @xmath460 if and only if there exist borel measurable bijections @xmath54 and @xmath107 such that @xmath450 and @xmath451 almost surely . * if @xmath452 and @xmath453 are strictly monotonic transformations on images of @xmath1 and @xmath2 , respectively , then @xmath461 .
now , the property [ renyi : scale].@xmath462 means that @xmath459 is invariant under only strictly monotonic transformations of random variables . using the @xmath0-norm
which is invariant under all borel measurable bijections , we define @xmath463 where the last equality follows from proposition [ props](3 ) .
since the @xmath0-norm shares many properties with the sobolev norm ( see proposition [ prop:*norm ] ) , the properties of @xmath464 are for the most part analogous to those of @xmath459 s .
main exceptions are that [ renyi:1].@xmath462[renyi : scale].@xmath462 are replaced back by [ renyi:1].[renyi : scale ] . 1 .
2 . @xmath467 3 .
@xmath468 if and only if @xmath1 and @xmath2 are independent .
[ * :1 ] @xmath469 if @xmath2 is completely dependent on @xmath1 or @xmath1 is completely dependent on @xmath2 .
[ * : scale ] if @xmath54 and @xmath107 are borel measurable bijective transformations , then we have @xmath470 6 . if @xmath471 is a sequence of pairs of continuous random variables with copulas @xmath472 and if @xmath473 , then @xmath474 1 .
follows from the fact that @xmath475 .
see proposition [ props ] .
2 . is clear from the definition of @xmath476 and the fact that @xmath477 $ ] .
the statement 3 . is a result of proposition [ props ] which says that @xmath478 if and only if @xmath479 .
4 . follows immediately from theorem [ thm : cdnorm1 ] . to prove 5 .
, let @xmath480 be borel measurable bijective transformations . then , @xmath1 and @xmath321 are mutually completely dependent , and so are @xmath2 and @xmath320 . thus @xmath481 and @xmath482
. therefore , the copulas @xmath483 and @xmath484 are invertible .
hence @xmath485 finally , 6 . can be proved via the inequality @xmath486 therefore , we have constructed a measure of dependence for continuous random variables which satisfies all of renyi s postulates except possibly the last condition [ renyi : rho ] .
the @xmath0-norm of a convex sum of a unit @xmath0-norm copula and the independence copula is computed .
corollary [ cor : l*r ] implies that there are many more copulas with unit @xmath0-norm , i.e. any copulas of the form @xmath490 where @xmath480 are borel measurable transformations . by the characterization of idempotent copulas in @xcite , all singular idempotent copulas are of this form and hence have unit @xmath0-norm .
00 brown , j.r .
doubly stochastic measures and markov operators , michigan math .
j. 12:367375 .
chou , s.h .
nguyen , t.t .
on frchet theorem in the set of measure preserving functions over the unit interval . international journal of mathematics and mathematical sciences 13:373378 .
chaidee , n. santiwipanont , t. sumetkijakan , s. ( 2012 ) .
denseness of patched mins in the sobolev norm , preprint .
darsow , w.f .
nguyen , b. olsen , e.t .
copulas and markov processes .
illinois j. math 36:600642 .
darsow , w.f .
olsen , e.t .
norms for copulas . international journal of mathematics and mathematical sciences 18:417436 .
darsow , w.f .
olsen , e.t .
characterization of idempotent 2-copulas .
note di matematica 30:147177 .
mikusinski , p. sherwood , h. taylor , m.d .
probabilistic interpretations of copulas and their convex sums . in : dallaglio ,
g. kotz , s. salinetti , g. ed .
, _ advances in probability distributions with given marginals : beyond the copulas . _ 67:95112 .
mikusinski , p. sherwood , h. taylor , m.d .
shuffles of min , stochastica 13:6174 nelsen , r.b .
( 2006 ) . _
an introduction to copulas , _
2@xmath491 ed .
springer verlag .
siburg , k.f .
stoimenov , p.a .
( 2008 ) . a scalar product for copulas .
journal of mathematical analysis and applications 344:429439 .
siburg , k.f .
stoimenov , p.a .
( 2009 ) . a measure of mutual complete dependence .
metrika 71 : 239251 .
sklar , m. ( 1959 ) .
fonctions de rpartition @xmath56 dimensions et leurs marges .
paris 8 : 229231 . | using a characterization of mutual complete dependence copulas , we show that , with respect to the sobolev norm , the mcd copulas can be approximated arbitrarily closed by shuffles of min .
this result is then used to obtain a characterization of generalized shuffles of copulas introduced by durante , sarkoci and sempi in terms of mcd copulas and the @xmath0-product discovered by darsow , nguyen and olsen .
since shuffles of a copula is the copula of the corresponding shuffles of the two continuous random variables , we define a new norm which is invariant under shuffling .
this norm gives rise to a new measure of dependence which shares many properties with the maximal correlation coefficient , the only measure of dependence that satisfies all of rnyi s postulates .
copulas , shuffles of min , measure - preserving , sobolev norm @xmath0-product , shuffles of copulas , measure of dependence 28a20 , 28a35 , 46b20 , 60a10 , 60b10 |
recently , as natural alternatives to general relativity , the scalar - tensor theories have been studied by many theoretical physicists . in these theories ,
the gravity is mediated not only by a tensor field but also by a scalar field .
also , such theories have been of interest as effective theories of the string theory at low energy scales @xcite .
several theoretical predictions in the scalar - tensor theories have been obtained ( see e.g. ref.@xcite@xmath0@xcite ) .
it has been found that a wide class of the scalar - tensor theories can pass all the experimental tests in weak gravitational fields .
however , it has also been found that the scalar - tensor theories show different aspects of the gravity in the strong gravitational fields in contrast to general relativity ( see e.g. ref.@xcite ) .
it has been shown numerically that nonperturbative effects in the scalar - tensor theories increase the maximum mass of an isolated system such as a neutron star [ 2 - 5 ] . in these works ,
numerical methods play an important role , and , in interpreting the numerical results , a static , spherically symmetric vacuum solution ( a spherical exterior solution ) is matched to the numerical solution .
the spherical exterior solution is an analytic exact solution , and , as far as the authors know , only few exact solutions have been known . in particular , the axisymmetric exterior solution corresponding to the kerr solution must be of significant interest .
motivated by this fact , we study series of the stationary solutions with asymptotic flatness properties in the einstein - maxwell - free scalar system because , as summarized in appendix a , they are locally equivalent with the exterior solutions in some class of the scalar - tensor theories of gravity .
the field equations are given by @xmath1 where @xmath2 is a energy - momentum tensor of the maxwell field ( see also appendix a ) .
in this paper we first classify spherical exterior solutions into two types of the solutions , an apparently black hole type ( abh ) solution and an apparently worm hole type ( awh ) solution .
we then clarify physical significance of the parameters contained in the solutions and show that the abh solution and the awh solution correspond to , respectively , the mass dominant case and the charge dominant case of the reissner - nordstrm solution .
second , we reduce the field equations for the axisymmetric exterior solutions and show that the reduced equations are partially the same as the ernst equations . as simple examples , we derive new series of static , axisymmetric exterior solutions , which will be referred to as scalar - tensor - weyl solutions .
we then show a non - trivial relation between the spherical exterior solutions and our new solutions .
finally , we study the local geometry of the exterior solutions by using the optical scalar equations and find some anomalous behaviors of the null geodesics .
in this section , we derive static , spherically symmetric exterior solutions with the electric field , which are hereafter referred to as spherical exterior solutions . in the just coordinate ,
the metric becomes @xcite : @xmath3 where @xmath4 and @xmath5 are functions of @xmath6 .
non - vanishing components of the electromagnetic tensor , @xmath7 , are @xmath8 .
the maxwell equation is reduced to @xmath9 where a prime denotes a derivative with respect to @xmath6 . with ( [ e3 ] )
one obtains @xmath10 where the integration constant , @xmath11 , is an electric charge .
the einstein equation and the field equation for @xmath12 become @xmath13 with ( [ e5 ] ) and ( [ e7 ] ) , one obtains @xmath14 one finds two types of the solutions to ( [ e9 ] ) , which are referred to as an _ apparently black hole type _ ( abh ) solution and an _ apparently worm hole type _ ( awh ) solution , respectively : @xmath15 where @xmath16 is an integration constant . when @xmath17 , the equation ( [ e8 ] ) is integrated as @xmath18 , \label{e14}\end{aligned}\ ] ] where @xmath19 and @xmath20 are integration constants [ 2 - 5 ] .
the equation @xmath21 is integrated as @xmath22 where @xmath23 is an integration constant , and @xmath24 is 1 in the abh solution and @xmath25 in the awh solution . with ( [ e5 ] ) , ( [ e6 ] ) and ( [ e7 ] )
, we obtain @xmath26 which is reduced to the following relation among the integration constants : @xmath27 since @xmath28 as @xmath29 in the asymptotically flat spacetime , with ( [ e16 ] ) , we obtain the following inequality for the awh solution : @xmath30 that is , the awh solution is _ charge dominant _ , and no awh solution exists when @xmath31 .
the metric function , @xmath4 , in the abh solution is obtained as follows : @xmath32 where @xmath33 in the limit , @xmath34 , the solution coincides with the previously known solution [ 2 - 5 ] . in the limit , @xmath35 , the solution is reduced to the following form : @xmath36^{-2}. \label{e21}\ ] ] when @xmath37 , i.e. , @xmath38 , the just coordinate , @xmath6 , is related to the schwarzschild coordinate , @xmath39 , as @xmath40 the metric function , @xmath4 , becomes @xmath41 \left[1-\left(\frac{b}{1-k}\right)\frac{1}{r}\right ] \equiv \left(1-\frac{r_{-}}{r}\right)\left(1-\frac{r_{+}}{r}\right ) \equiv 1-\frac{2m}{r}+\frac{q^2}{r^2 } , \label{e37}\ ] ] where @xmath42 with ( [ e41 ] ) , we obtain the following inequality : @xmath43 that is , the abh solution corresponds to the _ mass dominant _ ( @xmath44 )
reissner - nordstrm spacetime and has two event horizons , @xmath45 and @xmath46 , when @xmath37 . when @xmath47 , the just coordinate , @xmath6 , is related to the schwarzschild coordinate , @xmath39 , as @xmath48 which vanishes at @xmath49 , @xmath50 and @xmath51 .
moreover , we find that @xmath49 , @xmath52 and @xmath53 , correspond to , respectively , @xmath54 , @xmath55 and @xmath56 when @xmath37 and that the null surface , @xmath49 , becomes a singularity when @xmath47 . when @xmath57 , the metric function , @xmath4 , in the awh solution is obtained as @xmath58 , \label{e29}\ ] ] where the constant , @xmath59 , is defined by @xmath60=\frac{4q^2}{b^2}. \label{e30}\ ] ] as mentioned previously , the awh solution does not exist when @xmath31 . when @xmath61 , the awh solution becomes @xmath62 where @xmath63 when @xmath37 , the awh solution is reduced to the following : @xmath64 , \label{e46 } \\
\varphi & = & \varphi_0 , \label{e45 } \end{aligned}\ ] ] where @xmath65 note that ( [ e47 ] ) is equivalent with the following : @xmath66 then ( [ e46 ] ) becomes @xmath67 and the just coordinate , @xmath6 , is related to the schwarzschild coordinate , @xmath39 , as @xmath68 when we take @xmath69 , we have @xmath70 which is a negative mass solution and should be discarded as an unphysical exterior solution .
when we take @xmath71 , we have @xmath72 that is , the awh solution corresponds to the _ charge dominant _ ( @xmath73 ) reissner - nordstrm spacetime and has a naked timelike singularity , @xmath56 , when @xmath37 . when @xmath74 , the just coordinate , @xmath6 , is related to the schwarzschild coordinate , @xmath39 , as @xmath75 .
\label{e61}\ ] ] note that @xmath56 at @xmath76 , where @xmath77.\ ] ] the awh solution has a naked timelike singularity at @xmath76 .
that is , the awh solution does not have a worm hole , and the asymptotic region defined by @xmath78 does not exist .
in the stationary , axisymmetric spacetime , there exist two killing vectors , @xmath79 and @xmath80 .
it is shown that the energy - momentum tensor , @xmath81 , of the maxwell field satisfies the following relations @xcite : @xmath82}= \eta^{\alpha}t_{\alpha}^{\ [ \beta}\xi^{\gamma}\eta^{\delta]}= 0 .
\label{a01}\ ] ] moreover , one finds that the similar relations hold for @xmath83}\equiv 2\partial_{\mu}\varphi\partial_{\nu}\varphi$ ] .
accordingly , with the theorem 7.1.1 in ref.@xcite , the metric can be reduced to the following form : @xmath84 , \label{a1}\ ] ] where the metric functions , @xmath85 and @xmath86 , are functions of @xmath87 and @xmath88 . after long and complicated calculations , we find that the field equations are reduced to the following : @xmath89 , \label{a8 } \\
e^{2\psi}\nabla^2{\phi } & = & ( \nabla\phi)\cdot\left[(\nabla{\mathcal{e}})+2\bar{\phi}\nabla\phi\right ] , \label{a9 } \\
\gamma_{,1}&= & \rho\left[(\psi_{,1})^2-(\psi_{,2})^2\right ] -\frac{1}{4\rho}\left[(\omega_{,1})^2-(\omega_{,2})^2\right ] { { \rm e}}^{4\psi } \nonumber \\ & & -\rho(\phi_{,1}\bar{\phi}_{,1}-\phi_{,2}\bar{\phi}_{,2}){{\rm e}}^{-2\psi } + \rho\left[(\varphi_{,1})^2-(\varphi_{,2})^2\right ] , \label{a10 } \\ \gamma_{,2}&= & 2\rho\psi_{,1}\psi_{,2 }
-\frac{1}{2\rho}\omega_{,1}\omega_{,2}{{\rm e}}^{4\psi } -\rho(\phi_{,1}\bar{\phi}_{,2}+\phi_{,2}\bar{\phi}_{,1}){{\rm e}}^{-2\psi } + 2\rho\varphi_{,1}\varphi_{,2 } , \label{a11 } \\ \nabla^2\varphi & = & 0 , \label{a12}\end{aligned}\ ] ] where , for any functions , @xmath90 and @xmath91 , @xmath92 in ref.@xcite , one will find the explicit forms of the metric function , @xmath86 , and the maxwell field in terms of the complex potentials , @xmath93 and @xmath94 . note that the first , second and third equations are the same as those in general relativity and are referred to as _ the ernst equations _ of the einstein - maxwell system .
the remaining equations contain the scalar field contributions , and , as for the metric functions , effects of the scalar field only appear in @xmath4 .
in this section , we shall consider static , axisymmetric vacuum solutions . in this case ,
@xmath95 , and the field equations ( [ a7])@xmath0([a12 ] ) are reduced to the following : @xmath96 + \rho\left[(\varphi_{,1})^2-(\varphi_{,2})^2\right ] , \label{b3 } \\
\gamma_{,2}&= & 2\rho\psi_{,1}\psi_{,2 } + 2\rho\varphi_{,1}\varphi_{,2 } , \label{b4 } \\ \nabla^2\varphi & = & 0 , \label{b5}\end{aligned}\ ] ] where @xmath97 is a real function .
one finds that the ernst equation ( [ b2 ] ) is reduced to @xmath98 that is , @xmath12 and @xmath99 are harmonic functions .
we introduce _ oblate _ and _ prolate _ coordinates , @xmath100 , defined by @xmath101 where @xmath102 is a positive constant , and @xmath103 .
the cases , @xmath104 and @xmath105 , are referred to as _ oblate _ and _ prolate _ , respectively . the ernst equation
is then reduced to @xmath106 + \frac{\partial}{\partial y}\left[(1-y^2)\frac{\partial\psi}{\partial y}\right ] = 0 .
\label{b10}\ ] ] the similar equation holds for @xmath12 .
the simplest prolate solution for @xmath99 is given by @xmath107 where @xmath108 is an integration constant .
the similar solution for @xmath12 is given by @xmath109 where @xmath19 and @xmath110 are integration constants .
then the corresponding solution for @xmath4 becomes @xmath111 where @xmath112 for completeness , an explicit form of the metric is shown below : @xmath113 .
\label{b16}\end{aligned}\ ] ] note that , though @xmath99 and @xmath12 are @xmath114-independent , @xmath4 depends on both @xmath115 and @xmath114 .
the metric contains two parameters , @xmath108 and @xmath116 .
now we show that the metric is reduced to the previously known one when we take specific values of the parameters . when @xmath117 , i.e. , @xmath118 , the metric is reduced to one of the weyl series of the solutions in general relativity ( voorhees s prolate solution @xcite ) .
therefore , we refer to ( [ b16 ] ) as the scalar - tensor - weyl solution . voorhees s prolate solution contains the schwarzschild solution as a specific case , @xmath119 , and the schwarzschild coordinates are related to @xmath100 as @xmath120 where @xmath121 is a usual mass parameter .
the case , @xmath122 , is the most interesting one .
we introduce new parameters , @xmath16 , @xmath23 and @xmath20 , and new coordinates ( the just coordinates ) by @xmath123 and @xmath124 then we find that the solution coincides with the spherical exterior solution with a vanishing electric charge : @xmath125 moreover , when @xmath119 , i.e. , @xmath37 , the solution is reduced to schwarzschild s one .
the right hand sides in ( [ b3 ] ) and ( [ b4 ] ) are in the forms of the superposition of the contributions of @xmath99 and @xmath12 .
it is important to note that the superposition of the _ effectively non - spherical _
@xmath99-terms corresponding to voorhees s prolate solution with @xmath126 and the _ spherical _ configuration of @xmath12 reduces to the solution containing the spherical exterior solution . when the spherical scalar field configuration is added to the @xmath99-terms corresponding to voorhees s prolate solution with @xmath119 , namely the schwarzschild solution , we can not have spherical solutions unless @xmath37 . the relation among these solutions
is schematically shown in figure 1 .
the scalar - tensor - weyl solution generically has a singularity , @xmath127 , whose topology is @xmath128 , and qualitative features of the singularity may be similar to those in voorhees s prolate solution . when @xmath122 , the singularity , @xmath127 , becomes a point , which is given @xmath49 in
the just coordinate . moreover , when @xmath129 , the topology of @xmath49 , corresponding to the event horizon , @xmath130 , becomes @xmath128 .
the simplest oblate solution for @xmath99 is given by @xmath131 where @xmath108 is an integration constant .
the similar solution for @xmath12 is given by @xmath132 where @xmath19 and @xmath110 are integration constants .
then the corresponding solution for @xmath4 becomes @xmath133 where @xmath134 . for completeness , an explicit form of the metric
is shown below : @xmath135 .
\label{b45}\end{aligned}\ ] ] when @xmath117 , the metric is reduced to voorhees s oblate solution .
in contrast to the prolate case , the oblate solution is not reduced to the spherical exterior solutions for any @xmath108 and @xmath116 .
in this section , we examine local behaviors of irrotational null geodesics in the spherical exterior solution with a vanishing electric charge .
the metric is rewritten in the schwarzschild coordinate as @xmath136 where the mass function , @xmath137 , can be interpreted as a local energy @xcite .
one immediately finds that @xmath137 becomes negative when @xmath138 .
therefore , one may expect that there will be significant difference not only in the global causal structures but also in the local geometry , depending on whether @xmath16 is equal to @xmath23 or not .
however , we will find that it is not the case and that local behaviors of null geodesics significantly depend on whether @xmath16 is greater than @xmath139 or not . due to the conformally invariant nature of null geodesics
, we consider null geodesics in the spacetime with a metric , @xmath140 ( see appendix a ) . let @xmath141 be null geodesic associated with @xmath142 with the affine parameter , @xmath143 .
hereafter , a hat denotes geometrical quantities associated with @xmath142 , and the corresponding geometrical quatities associated with @xmath144 are denoted without a hat
. then the null geodesics associated with @xmath144 are given as @xmath145 since the local geometrical nature of the spacetime is described by the riemann curvature , it is important to examine optical scalars of the irrotational null geodesic congruence defined by sachs @xcite .
let @xmath146 be a null tetrad such that @xmath147 then the optical scalars , namely , the expansion , @xmath148 , and the complex shear , @xmath102 , are defined as @xmath149 the optical scalar equations are also conformally invariant , and we define the following quantities associated with @xmath144 : @xmath150 then the optical scalar equations become @xcite @xmath151 where @xmath152 is the weyl curvature .
first , we examine circular orbits of null geodesics in the just coordinate . without loss of generality
, we consider null geodesics on the @xmath153 plane .
then the geodesic equation is reduced to @xmath154 where a dot denotes a derivative with respect to the affine parameter , @xmath5 , and @xmath155 and @xmath156 are integration constants .
the circular orbits are determined by the conditions that @xmath157 and @xmath158 .
we find that the circular orbit is obtained as @xmath159 for @xmath160 , and that , surprisingly , no circular orbit exists for @xmath161 in contrast to the case in the schwarzschild spacetime . in summary
, we have found that local behaviors of null geodesics significantly depend on whether @xmath16 is greater than @xmath139 or not .
when @xmath160 , the behaviors are similar to those in the schwarzschild spacetime .
that is , the circular orbit exists , and the weyl driving term , @xmath163 , is strictly positive . when @xmath174 , on the contrary , there is no circular orbit , and @xmath163 changes its sign at @xmath178 .
these behaviors of null geodesics can be compared with those in the reissner - nordstrm spacetime : @xmath179 the ricci and weyl driving terms are calculated as @xmath180 one immediately finds that @xmath163 vanished at @xmath181 . however , when @xmath44 , @xmath182 is inside the event horizon , @xmath183 .
when @xmath184 , two circular orbits exist : @xmath185 moreover , when @xmath44 , one finds that @xmath186 , where @xmath187 .
a particular case that @xmath188 occurs only when @xmath189 , and one finds that @xmath190 and that @xmath191 .
general solutions to @xmath192 with @xmath193 can be obtained by dyer s formula ( appendix b.1 ) , however , we have only examined quantitative behaviors of the ricci and weyl driving terms , @xmath167 and @xmath163 .
further quantitative studies of the solutions and their astronomical applications , especially the implication to gravitational lens effects , will be discussed in the forthcoming paper .
it is difficult to study generic null geodesics in the scalar - tensor - weyl solution .
therefore , we shall examine specific null geodesics on the @xmath194 plane .
when @xmath194 , the geodesic equation is reduced to @xmath195 where @xmath196 is a function of @xmath115 defined by @xmath197 a dot denotes a derivative with respect to the affine parameter , @xmath5 , and @xmath155 and @xmath156 are integration constants .
the circular orbits are determined by the conditions that @xmath198 and @xmath199 .
we find that these conditions are reduced to @xmath200 .
since @xmath115 should be larger than unity , the circular orbit exists when @xmath201 .
we evaluate the ricci and weyl deriving terms , @xmath167 and @xmath163 , ( see appendix b.2 ) and find that @xmath202 \\ & = & -\dfrac{\delta^2-\delta^2}{(x^2 - 1)^2}(k^1)^2 , \\ f & = & \dfrac{1}{\sigma^2}\dfrac{1}{x^3(x^2 - 1)}\left(\dfrac{x^2}{x^2 - 1 } \right)^{\delta^2 } \times \\ & & \times \left[\delta(\delta^2 - 1)+\dfrac{(h/\sigma)^2 } { x^2 - 1 } \left(\dfrac{x-1}{x+1}\right)^{2\delta } \left\{\delta(\delta^2 - 1)-(\delta^2 + 2\delta^2)x+3\delta x^2\right\ } \right ] , \end{array } \label{n2}\ ] ] where @xmath169 is an impact parameter .
one finds that they are reduced to ( [ ge10 ] ) when @xmath203 .
we show @xmath204 and @xmath163 in figure 2 .
we numerically find that the qualitative features of @xmath167 and @xmath163 are the same as those in the spherical case ( @xmath122 ) and that the significant qualitative changes in @xmath204 and @xmath163 are seen depending on whether @xmath205 or not .
after summarizing the spherical exterior solutions in the scalar - tensor theories of gravity , we have derived a new series of the axisymmetric exterior solutions , which are reduced to voorhees s solutions in general relativity .
we have found that the prolate solution , namely the scalar - tensor - weyl solution , contains the spherical exterior solution as a special case in which we take some non - trivial choice of the parameters in the solution .
we also tried to find exterior solutions corresponding to the kerr solution in general relativity and have not yet succeeded in it .
note that the generic scalar - tensor - weyl solution is not reduced to the schwarzschild solution in the limit , @xmath206 .
therefore , we guess that the kerr - like exterior solution must be reduced to the tomimatsu - sato - like solution in the limit , @xmath206 , where , by the term , the kerr - like exterior solution , we mean the exterior solution such that it coincides with the kerr solution when @xmath207 and that it is reduced to the spherical exterior solution when the kerr parameter ( the angular momentum of the matter ) vanishes .
an example of the solution with the first property is obtained by using the metric function , @xmath99 , corresponding to the kerr solution .
however , the resulting solution does not satisfy the second property .
we think that generating methods of the exact solutions ( e.g. the newman - janis algorithm @xcite ) may play an important role in obtaining further exterior solutions .
it is important to note that one have to specify an explicit form of the coupling function , @xmath208 , in order to obtain the complete information , especially the global geometry , of the exterior solutions in the scalar - tensor theories .
we have not given any discussion on @xmath208 , and , instead , we have examined local geometrical properties of the solutions by taking account of the conformally invariant nature of null geodesics .
we have evaluated the ricci and weyl deriving terms , @xmath167 and @xmath163 , in the scalar - tensor - weyl solution and have found that local behaviors of null geodesics significantly depend on whether the parameter , @xmath108 , which corresponds to @xmath209 in the spherical exterior solution with @xmath31 , is greater than @xmath210 or not .
when @xmath201 , the behaviors are similar to those in the schwarzschild spacetime .
that is , the circular orbit exists , and the weyl driving term , @xmath163 , is strictly positive .
when @xmath205 , on the contrary , there is no circular orbit , and @xmath163 changes its sign at @xmath178 . in the later case , one may expect to observe a gravitational lensing anomaly for images of distant sources appearing near @xmath178 .
a ratio , @xmath211 is a measure of the image deformation . in the schwarzschild solution , @xmath212 is a positive , decreasing function of @xmath91 . however , in the spherical exterior solution with @xmath205 , @xmath212 may become a negative increasing function of @xmath91 near @xmath178 .
further quantitative studies of the solutions , @xmath213 and @xmath214 , and their astronomical applications , especially the implication to gravitational lens effects including their observational detectability , will be discussed in the forthcoming paper .
we shall consider the simplest scalar - tensor theory [ 2 - 5,14 ] . in this theory ,
gravitational interactions are mediated by a tensor field , @xmath215 , and a scalar field , @xmath216 .
hereafter , a symbol , @xmath217 , denotes quantities or derivatives associated with @xmath218 .
an action of the theory is the following : @xmath219\sqrt{-\hat{g}}d^4x + s_{\mbox{\tiny matter}}[\hat{\psi}_{\mbox{\tiny m}},\hat{g}_{\mu\nu } ] , \label{ea5}\ ] ] where @xmath220 is a dimensionless arbitrary function of @xmath221 , @xmath222 represents matter fields , and @xmath223 is an action of the matter fields .
the scalar field , @xmath221 , plays a role of an effective gravitational constant as @xmath224 .
varying the action by the tensor field , @xmath218 , and the scalar field , @xmath221 , yields , respectively , the following field equations : @xmath225 now we perform the following conformal transformation to the unphysical frame ( the einstein frame ) with the metric , @xmath144 : @xmath226 such that @xmath227 where @xmath228 is a bare gravitational constant , and @xmath208 is referred to as a coupling function .
then the action is rewritten as @xmath229 , \label{ea10}\ ] ] where the scalar field , @xmath12 , is defined by @xmath230 varying the action by @xmath231 and @xmath12 yields , respectively , @xmath232 where @xmath233 represents a energy - momentum tensor with respect to @xmath231 defined by @xmath234 } { \delta g_{\mu\nu}}=a^6(\varphi)\hat{t}^{\mu\nu}. \label{ea14}\ ] ] the conservation law of @xmath233 is given by @xmath235 for the maxwell field , one has @xmath236 further properties of the maxwell field under the conformal transformation are found in ref.@xcite . in this paper , we adopt the unit , @xmath237 .
in this appendix , we summarize the analytic results obtained by dyer @xcite . a metric of the static , spherically symmetric spacetime is given as @xmath238 where @xmath239 , @xmath240 and @xmath241 are functions of @xmath39 .
the null tangent vector , @xmath242 , to a null geodesic in this spacetime is obtained as @xmath243 where a constant , @xmath91 , is an impact parameter , and we assume that the geodesic is on the @xmath153 plane without loss of generality .
the complex null vector , @xmath244 , is obtained as @xmath245,~~t^2=\frac{i}{\sqrt{2}}{{\rm e}}^{-b},\cr & t^1=\frac{{{\rm e}}^{-a}}{2\sqrt{2}}\left[\frac{1}{h}(s_+^2+s_-^2){{\rm e}}^{b } + 2hs_+s_-\right],~~t^3=\frac{hh}{\sqrt{2}}{{\rm e}}^{-2b},\cr & s_\pm=\sqrt{{{\rm e}}^{-c}\pm h{{\rm e}}^{-b}},~~ h=-\frac{1}{h}\int^r\frac{b'}{s_+s_-}{{\rm e}}^{b - c}dr , } \label{gap3}\ ] ] where a prime denotes a derivative with respect to @xmath39 .
then the ricci and weyl driving terms , @xmath167 and @xmath163 , in @xmath162 are evaluated as @xmath246 since @xmath163 is real , one can let @xmath102 be real without loss of generality .
dyer has introduced the new optical scalars , @xmath164 , defined by @xmath247 and has obtained the following equations : @xmath248 dyer has obtained general solutions to @xmath249 as @xmath250^{\frac{3}{2}}}dr + d_+^0\right\ } , \cr & c_-=c_-^0{{\rm e}}^{b}\sin\left(h \int^r \frac{{{\rm e}}^{a+c}}{\sqrt{{{\rm e}}^{2b}-h^2{{\rm e}}^{2c}}}dr + d_-^0\right ) , \cr } \label{gap7}\ ] ] where @xmath251 and @xmath252 are integration constants . a metric of the static , axisymmetric spacetime is given by @xmath253 where metric functions , @xmath254 , @xmath59 , @xmath4 and @xmath255 , are functions of @xmath115 and @xmath114 .
the tangent vector , @xmath242 , to a null geodesic on the @xmath194 plate is obtained as @xmath256 where a constant , @xmath91 , is an impact parameter .
the complex null vector , @xmath244 , is obtained as @xmath257{{\rm e}}^{-\alpha},~~ t^2=\frac{i}{\sqrt{2}}{{\rm e}}^{-\gamma } , \cr & t^1=\frac{1}{2\sqrt{2}}\left[\frac{1}{h}(s_{+}^2+s_{-}^2){{\rm e}}^{\mu } + 2hs_{+}s_{-}\right]{{\rm e}}^{-\beta},~~ t^3=\frac{1}{\sqrt{2}}hh{{\rm e}}^{-2\mu},\cr & s_{\pm } = \sqrt{{{\rm e}}^{-\alpha}\pm h{{\rm e}}^{-\mu } } , ~~ h=-\frac{1}{h}\int^x \frac{{{\rm e}}^{\mu-\alpha}}{s_{+}s_{- } } \frac{\partial\gamma}{\partial x}dx.\cr } \ ] ] 99 m. b. green , j. h. schwartz and e. witten , _ superstring theory vols .
1,2 _ , ( cambridge university press , cambridge , 1987 ) . t. damour and g. esposito - farse , class
@xmath258 , 2093 ( 1992 ) . t. damour and g. esposito - farse , phys . rev .
@xmath259 , 2220 ( 1993 ) .
t. damour and g. esposito - farse , phys .
@xmath260 , 1474 ( 1996 ) .
t. tsuchida , g. kawamura and k. watanabe , prog . theor . phys . , @xmath261 , in press , ( 1998 ) . b. carter , j. math .
phys . , @xmath262 , 70 , ( 1969 ) .
r. m. wald , _ general relativity _
, ( the university of chicago press , chicago and london , 1984 ) .
s. chandrasekhar , _ the mathematical theory of black holes _ ( oxford university press , new york , 1983 )
. b. voorhees , phys .
, @xmath263 , 2119 , ( 1970 ) .
s. hawking , j. math .
phys . , @xmath258 , 598 , ( 1968 ) .
r. k. sachs , proc .
london , @xmath264 , 309 , ( 1961 ) .
c. c. dyer , mon . not .
soc . , @xmath265 , 231 , ( 1977 ) .
e. t. newman and a. i. janis , j. math .
phys . , @xmath266 , 915 , ( 1965 ) .
c. brans and r. h. dicke , phys .
@xmath267 , 925 ( 1962 ) . | we study series of the stationary solutions with asymptotic flatness properties in the einstein - maxwell - free scalar system because they are locally equivalent with the exterior solutions in some class of the scalar - tensor theories of gravity .
first , we classify spherical exterior solutions into two types of the solutions , an apparently black hole type solution and an apparently worm hole type solution .
the solutions contain three parameters , and we clarify their physical significance .
second , we reduce the field equations for the axisymmetric exterior solutions . we find that the reduced equations are partially the same as the ernst equations . as simple examples ,
we derive new series of the static , axisymmetric exterior solutions , which correspond to voorhees s solutions .
we then show a non - trivial relation between the spherical exterior solutions and our new solutions .
finally , since null geodesics have conformally invariant properties , we study the local geometry of the exterior solutions by using the optical scalar equations and find some anomalous behaviors of the null geodesics . |
molecular modeling and simulation is a technology central to many areas of research in academia and industry . with the advance of computing power ,
the scope of application scenarios for molecular simulation is widening , both in terms of complexity of a given simulation and in terms of high throughput .
nowadays , e.g. the predictive simulation of entire phase equilibrium diagrams has become feasible .
however , in order to rely on simulation results , the methodology needs to be sound and the implementation must be thoroughly verified . in its first release @xcite
, we have introduced the molecular simulation tool @xmath02 .
results from @xmath02 have been verified and the implementation was found to be robust and efficient . as described in section [ hybrid ] , in version 2.0 of the simulation tool @xmath02 the existing molecular dynamics ( md )
mpi parallelization was hybridized with openmp , leading to an improved performance on multi - core processors .
furthermore , the new release offers a wider scope of accessible properties .
in particular , @xmath02 was extended to calculate massieu potential derivatives in a systematic manner , cf .
section [ massieu ] .
this augments the range of sampled properties significantly and , as was demonstrated in @xcite , it allows to straightforwardly develop competitive fundamental equations of state from a combination of experimental vle data and molecular simulation results .
lastly , besides being now capable of simulating ionic substances , the time and memory demand for calculating transport properties was reduced significantly ( section [ algo ] ) .
@xmath02 is freely available as an open source code for academic users at www.ms-2.de .
the molecular simulation tool @xmath02 focuses on thermodynamic properties of homogeneous fluids .
therefore , systems investigated with @xmath02 typically contain on the order of @xmath1 molecules . while for monte carlo simulations a perfect scaling behavior up to large numbers of cores can be trivially achieved , md domain decomposition
the de facto standard for highly scalable md is not feasible for such system sizes , because the cut - off radius is in the same range as half the edge length of the simulation volume .
this excludes domain decomposition and limits the scalability of the mpi parallelization .
the present release of @xmath02 features an openmp parallelization , which was hybridized with mpi . at the point where mpi communication becomes a bottleneck
, a single process still has enough load to distribute to multiple threads , improving scalability .
three parts of @xmath02 were parallelized with openmp : the interaction partner search , the energy and the force calculations .
all openmp parallel regions rely on loop parallelism , as the compute intensive parts of the algorithm all feature a loop over the molecules . in the force calculation ,
race conditions need to be considered , because every calculated force is written to both interacting molecules . introducing atomic updates or critical sections
leads to massive overheads .
instead , it is more efficient to assign forces from individual interactions to the elements of a list ( or an array ) which is subsequently summed up .
the same holds true for torques . in figure [ hybrid ] the speed - up of hybrid mpi / openmp vs. pure mpi is plotted for 2048 cores , varying the number of threads per mpi process and the number of molecules in the simulation volume . as can be seen , using 2 to 4 threads per mpi process delivers a speed - up of around 20% for 2048 cores .
the evaluation of the hybrid parallelization algorithm was performed on a cray xe6 supercomputer at the high performance computing center in stuttgart , which has an overall peak performance of one pflops .
it consists of 3552 nodes , each equipped with two amd opteron 6276 ( interlagos ) processors .
each processor has 16 cores , sharing eight fpus ( floating point units ) .
nodes are equipped with 32 gb ram and are interconnected by a high - speed cray gemini network .
additional runtime performance comparisons with the simulation tool _ gromacs _ @xcite are listed in table [ tab1 ] .
cccccc cores & threads & @xmath2 & gromacs / s & @xmath02(rf ) / s & @xmath02(ew ) / s + 8 & 8 mpi & 500 & 164 & 416 & 785 + 8 & 8 mpi & 1000 & 299 & 874 & 1607 + 8 & 8 mpi & 2000 & 1284 & 4461 & 6777 + 16 & 16
mpi & 500 & 95 & 233 & 415 + 16 & 16 mpi & 1000 & 166 & 477 & 848 + 16 & 16 mpi & 2000 & 678 & 2298 & 3506 + 32 & 32 mpi & 500 & 62 & 152 & 245 + 32 & 32 mpi & 1000 & 106 & 296 & 487 + 32 & 32 mpi & 2000 & 361 & 1286 & 1898 + 64 & 64 mpi & 500 & 40 & 119 & 166 + 64 & 64 mpi & 1000 & 65 & 228 & 324 + 64 & 64 mpi & 2000 & 220 & 814 & 1261 + 128 & 128 mpi & 500 & 38 & 105 & 131 + 128 & 128 mpi & 1000 & 51 & 197 & 247 + 128 & 128 mpi & 2000 & 147 & 557 & 727 + 8 & 1 mpi , 8 omp / mpi & 500 & 167 & 483 & + 8 & 1 mpi , 8 omp / mpi & 1000 & 323 & 975 & + 8 & 1 mpi , 8 omp / mpi & 2000 & 1416 & 4831 & + 16 & 2 mpi , 8 omp / mpi & 500 & 105 & 253 & + 16 & 2 mpi , 8 omp / mpi & 1000 & 186 & 517 & + 16 & 2 mpi , 8 omp / mpi & 2000 & 763 & 2514 & + 32 & 4 mpi , 8 omp / mpi & 500 & 75 & 167 & + 32 & 4 mpi , 8 omp / mpi & 1000 & 121 & 316 & + 32 & 4 mpi , 8 omp / mpi & 2000 & 418 & 1362 & + 64 & 8 mpi , 8 omp / mpi & 500 & 60 & 119 & + 64 & 8 mpi , 8 omp / mpi & 1000 & 92 & 217 & + 64 & 8 mpi , 8 omp / mpi & 2000 & 261 & 785 & + 128 & 16 mpi , 8 omp / mpi & 500 & 49 & 101 & + 128 & 16 mpi , 8 omp / mpi & 1000 & 74 & 172 & + 128 & 16 mpi , 8 omp / mpi & 2000 & 170 & 496 & + + + + [ tab1 ]
@xmath02 version 2.0 features evaluating free energy derivatives in a systematic manner , thus greatly extending the thermodynamic property types that can be sampled in single simulation runs .
the approach is based on the fact that the fundamental equation of state contains the complete thermodynamic information about a system , which can be expressed in terms of various thermodynamic potentials @xcite , e.g. internal energy @xmath3 , enthalpy @xmath4 , helmholtz free energy @xmath5 or gibbs free energy @xmath6 , with number of particles @xmath2 , volume @xmath7 , pressure @xmath8 , temperature @xmath9 and entropy @xmath10 .
these representations are equivalent in the sense that any other thermodynamic property is essentially a combination of derivatives of the chosen form with respect to its independent variables .
the form @xmath11 , known as the massieu potential , is preferred in molecular simulations due to practical reasons @xcite .
the statistical mechanical formalism of lustig allows for the simultaneous sampling of any @xmath12 in a single @xmath13 ensemble simulation for a given state point @xcite @xmath14 where @xmath15 is the gas constant , @xmath16 and @xmath17 .
@xmath18 can be separated into an ideal part @xmath19 and a residual part @xmath12 @xcite .
the calculation of the residual part is the target of molecular simulation and the derivatives @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26 and @xmath27 were implemented in @xmath02 for @xmath13 ensemble simulations .
the ideal part can be obtained by independent methods , e.g. from spectroscopic data or ab initio calculations .
however , it can be shown that for any @xmath28 , where @xmath29 , the ideal part is either zero or depends exclusively on the density , thus it is known by default @xcite .
note that the calculation of @xmath30 still requires additional concepts such as thermodynamic integration or particle insertion methods . from the first five derivatives @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 every measurable thermodynamic property can be expressed ( see the supplementary material for a list of properties ) with the exception of phase equilibria .
a detailed description of the implementation is in the supplementary material , here , only an overview is provided .
+ the calculation of the derivatives up to the order of @xmath36 requires the explicit mathematical expression of @xmath37 and @xmath38 with respect to the applied molecular interaction pair potential and has to be determined analytically beforehand @xcite . the general formula for @xmath39 can be found in ref .
@xcite . for common molecular interaction pair potentials , like the lennard - jones potential @xcite , describing repulsive and dispersive interactions , or coulomb s law , describing electrostatic interactions between point charges ,
the analytical formulas for @xmath37 and @xmath38 can be obtained straightforwardly .
+ as molecular simulation is currently limited to operate with considerably fewer particles than real systems , the effect of the small system size thus has to be counter - balanced with a contribution to @xmath40 and @xmath39 called long range correction ( lrc ) @xcite .
the mathematical form of the lrc depends on the molecular interaction potential and the cut - off method ( site - site or center - of - mass cut - off mode ) applied . for the lennard - jones potential ,
the lrc scheme was well described in the literature for both the site - site @xcite and the center of mass cut - off mode @xcite .
the reaction field method @xcite was the default choice in the preceding version of @xmath02 for the lrc of electrostatic interactions modelled by considering charge distributions on molecules .
the usual implementation of the reaction field method combines the explicit and the lrc part in a single pair potential @xcite from which @xmath39 ( including the lrc contribution ) is directly obtainable .
however , practical applications show that the electrostatic lrc of @xmath37 and @xmath38 can be neglected in case of systems for which the reaction field method is an appropriate choice .
e.g. , the contribution of the electrostatic lrc for a liquid system ( @xmath41 k and @xmath42 mol / l ) consisting of only 200 water and 50 methanol molecules with a very short cut - off radius of @xmath43 of the edge length of the simulation volume is still @xmath44 for both @xmath37 and @xmath38 .
the supplementary material contains detailed elaborations on the lrc for the lennard - jones potential .
[ [ transport - property - calculations ] ] transport property calculations + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in @xmath02 , transport properties are determined via equilibrium md simulations by means of the green - kubo formalism @xcite .
this formalism offers a direct relationship between transport coefficients and the time integral of the autocorrelation function of the corresponding fluxes .
an extended time step was defined for the calculation of the fluxes , the autocorrelation functions and their integrals .
the extended time step is @xmath45 times longer than the specified md time step , where @xmath45 is a user defined variable .
the autocorrelation functions are hence evaluated in every @xmath45-th md time step . as a consequence ,
the memory demand for the autocorrelation functions was reduced and the restart files , which contain the current state of the autocorrelation functions and time integrals , become accordingly smaller . in addition , the overall computing time of the md simulation was reduced significantly .
+ [ [ ewald - summation ] ] ewald summation + + + + + + + + + + + + + + + ewald summation @xcite was implemented for the calculation of electrostatic interactions between point charges .
it extends the applicability of @xmath02 to thermodynamic properties of e.g. ions in solutions . in ewald summation ,
the electrostatic interactions according to coulomb s law are divided into two contributions : short - range and long - range .
the short - range term includes all charge - charge interactions at distances smaller than the cut - off radius .
the remaining contribution is calculated in fourier space and only the final value is transformed back into real space .
this allows for an efficient calculation of the long - range interactions between the charges .
the algorithm is well described in literature . currently , some of the new features , the calculation of massieu potential derivatives and hybrid mpi & openmp parallelization for md , are not available together with ewald summation .
[ [ radial - distribution - function ] ] radial distribution function + + + + + + + + + + + + + + + + + + + + + + + + + + + + the radial distribution function ( rdf ) @xmath46 is a measure for the microscopic structure of matter .
it is defined by the local number density around a given position within a molecule @xmath47 in relation to the overall number density @xmath48 @xmath49 therein , @xmath50 is the differential number of molecules in a spherical shell volume element @xmath51 , which has the width @xmath52 and is located at the distance @xmath53 from the regarded position .
@xmath46 can be evaluated for every molecule of a given species . in the present release of @xmath02 , the rdf can be calculated during md simulation
runs for pure components and mixtures on the fly .
the rdf is sampled between all lj sites . in order to evaluate rdfs for arbitrary positions ,
say point charge sites , superimposed dummy lj sites with the parameters @xmath54 have to be introduced in the potential model file by the user .
[ [ electric - conductivity ] ] electric conductivity + + + + + + + + + + + + + + + + + + + + + the evaluation of the electric conductivity @xmath55 was implemented in @xmath02 version 2.0 , being a measure for the flow of ions in solution .
the green - kubo formalism @xcite offers a direct relationship between @xmath55 and the time - autocorrelation function of the electric current flux @xmath56
@xcite @xmath57 where @xmath58 is boltzmann s constant .
the electric current flux is defined by the charge @xmath59 of ion @xmath60 and its velocity vector @xmath61 according to @xmath62 where @xmath63 is the number of molecules of component @xmath64 in solution .
note that only the ions in the solution have to be considered , not the electro - neutral molecules . for better statistics ,
@xmath55 is sampled over all independent spatial elements of @xmath56 .
[ [ thermal - conductivity - of - mixtures ] ] thermal conductivity of mixtures + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the previous version of @xmath65 the determination of the thermal conductivity by means of the green - kubo formalism was implemented for pure substances only . in the present release
, the calculation of the thermal conductivity was extended to multi - component mixtures .
the thermal conductivity @xmath66 is given by the autocorrelation function of the elements of the microscopic heat flow @xmath67 @xmath68 in mixtures , energy transport and diffusion occur in a coupled manner , thus , the heat flow for a mixture of @xmath45 components is given by @xcite @xmath69\cdot \mathbf{v}_i^k \\ & -&\frac{1}{2}\sum_{i=1}^{n } \sum_{j=1}^{n } \sum_{k=1}^{n_i } \sum_{l \neq k}^{n_j } \bm{r}_{ij}^{kl}\cdot \big ( \mathbf{v}_i^k \cdot \frac{\partial u\left(r_{ij}^{kl}\right)}{\partial \bm{r}_{ij}^{kl}}+\mathbf{w}_i^k \mathbf{\gamma}_{ij}^{kl}\big)- \sum_{i=1}^{n } h_{i}\sum_{k=1}^{n_i } \mathbf{v}_{i}^{k}\mbox { , } \label{heatflow } \end{aligned}\ ] ] where @xmath70 is the angular velocity vector of molecule @xmath60 of component @xmath71 and @xmath72 its matrix of angular momentum of inertia .
@xmath73 is the intermolecular potential energy and @xmath74 is the torque due to the interaction of molecules @xmath60 and @xmath75 .
the indices @xmath71 and @xmath64 denote the components of the mixture .
@xmath76 is the partial molar enthalpy .
it has to be specified as an input in the @xmath65 parameter file and can be calculated from @xmath77 simulations .
[ [ residence - time ] ] residence time + + + + + + + + + + + + + + the residence time @xmath78 defines the average time span that a molecule of component @xmath64 remains within a given distance @xmath79 around a specific molecule @xmath71 .
it is given by the autocorrelation function @xmath80 where @xmath81 is the time , @xmath82 the solvation number around molecule @xmath71 at @xmath83 and @xmath84 is the heaviside function , which yields unity , if the two molecules are within the given distance , and zero otherwise . following the proposal of impey et al .
@xcite , the residence time explicitly allows for short time periods during which the distance between the two molecules exceeds @xmath79 .
also , the solvation number @xmath85 can be evaluated on the fly @xmath86 where @xmath87 is the number density of component @xmath64 and @xmath88 is the distance up to which the solvation number is calculated .
the authors gratefully acknowledge financial support by the bmbf `` 01ih13005a skasim : skalierbare hpc - software fr molekulare simulationen in der chemischen industrie '' and computational support by the high performance computing center stuttgart ( hlrs ) under the grant mmhbf2 .
the present research was conducted under the auspices of the boltzmann - zuse society for computational molecular engineering ( bzs ) .
19 natexlab#1#1[1]`#1 ` [ 2]#2 [ 1]#1 [ 1]http://dx.doi.org/#1 [ ] [ 1]pmid:#1 [ ] [ 2]#2 , , , , , , , , , : ( ) .
, , , , : ( ) .
( ) . : ( ) . : ( ) . , : , , . , : , , . , : , , . , : ( ) . : ( ) . , : ( ) . : ( ) .
: , , . , : , , . , : ( ) . , , : ( ) | a new version release ( 2.0 ) of the molecular simulation tool _ ms_2 [ s. deublein et al . , comput .
phys .
commun . 182
( 2011 ) 2350 ] is presented .
version 2.0 of _ ms_2 features a hybrid parallelization based on mpi and openmp for molecular dynamics simulation to achieve higher scalability .
furthermore , the formalism by lustig [ r. lustig , mol .
phys . 110
( 2012 ) 3041 ] is implemented , allowing for a systematic sampling of massieu potential derivatives in a single simulation run .
moreover , the green - kubo formalism is extended for the sampling of the electric conductivity and the residence time . to remove the restriction of the preceding version to electro - neutral molecules
, ewald summation is implemented to consider ionic long range interactions . finally , the sampling of the radial distribution function is added . |
this note is a write - up of talks given at the workshop `` vertex operator algebras in mathematics and physics '' at the fields institute in toronto in october 2000 and at some other occasions before .
its purpose is to explain how results in low dimensional quantum field theory obtained from different perspectives and with different motivations can be put into a uniform picture by generalizing the notions of discriminant forms and genera from integral quadratic forms to vertex operator algebras , indicating a rich underlying arithmetical theory .
the relation between two - dimensional conformal field theory and three - dimensional topological quantum field theory has been quite known from the beginning .
for example , witten relates in his well - known paper @xcite the wess - zumino - witten models with chern - simons theories and the jones polynomial of knots .
the basic idea is to use the monodromy properties of correlation functions on surfaces to obtain invariants for knots and three - dimensional manifolds .
being not overly precise , one can say that a quantum field theory is a functor from some bordism category of oriented manifolds to some category of linear spaces satisfying some basic gluing axioms .
@xmath0 @xmath1 @xmath2 @xmath3 @xmath4 for two - dimensional conformal field theories , one takes two - dimensional surfaces with a conformal structure on it and one is forced to take infinite dimensional linear spaces .
using instead riemannian surfaces ( which have a naturally defined conformal structure ) , one reaches the more restricted but essential notion of chiral algebras . for three - dimensional topological quantum field theories ( 3d - tqfts ) ,
one takes three - dimensional topological manifolds ( together with extra structures like framed links ) and only finite dimensional linear spaces .
it took some time before precise and efficient mathematical formulations for both structures , originating partly from physics , emerged .
chiral algebras have been axiomatized by the algebraic structure of _ vertex algebras _ ; see @xcite for some approaches .
3d - tqfts can be constructed from _
modular braided tensor categories _ ; see @xcite . in this paper
, we will restrict ourselves mainly to vertex operator algebras ( voas ) ; these are vertex algebras which have a virasoro element and we assume that they are non - negatively graded ; we consider only simple and rational voas and we make some further strong assumptions .
there are many different kinds of examples of voas known , but it is quite unclear yet how a general structure theory for them may look like . an important class of voas can be constructed from positive definite even integral lattices . for integral lattices ,
there is a deeply developed structure theory available leading to invariants like rational equivalence , the genus , or the spinor genus .
the first purpose of this paper is to explain how the notion of the _ genus _ can be generalized in a natural way from lattices to voas by using modular braided tensor categories so that the definition is compatible with the notion of the genus for lattices .
secondly , we show that new voas can be constructed from given ones only by using the associated modular braided tensor category , generalizing thereby the well - known discriminant technique of quadratic form theory ; cf .
@xcite .
it should be mentioned that there are papers , which do not distinguish between voas and the associated modular braided tensor category . usually , the authors make assumption [ hauptvermutung ] given below and work in the setting of 3d - tqfts .
the example of the lattice voas associated to the two positive definite even unimodular lattices @xmath5 and @xmath6 in dimension @xmath7 @xcite shows that there are different voas which can lead to the same 3d - tqft and which have the same central charge . the paper is organized as follows . in the second section ,
some basic definitions and results from the theory of integral quadratic forms are recalled . in section three
, we define the genus of a vertex operator algebra and try to formulate theorems or at least conjectures analogous to the ones known for quadratic forms . in the final section ,
we reformulate the extension problem for vertex operator algebras in terms of the associated modular braided tensor category as already outlined in my paper @xcite .
this extension problem can be considered as a kind of problem in coding theory over 3d - tqfts .
the solution for voas defining an abelian intertwining algebra is given .
also , complete results for self - dual framed voas or , equivalently , self - dual codes over the 3d - tqft associated to the ising model , are presented .
there are many open problems about the structure of vertex algebras compared to what is known for quadratic forms .
but the theory of quadratic forms is more than two hundred years old .
it seems to me that from the fifteen years old theory of vertex algebras we can see only the tip of the iceberg of such a structure theory .
i like to thank the organizers of the conference for inviting me to the fields institute and for the possibility to present these ideas .
i am grateful to markus rosellen for his comments on an early version of the paper and for discussions about abelian intertwining algebras as well as to geoffrey mason for reminding me of some references .
i also like to thank jim lepowsky and the referee for their many valuable suggestions .
finally , my thanks goes to supun koranasophonpun for proofreading the manuscript .
we collect some results from the arithmetic of integral quadratic forms .
the theory of genera for binary forms was developed by gau in @xcite . instead of the more classical notion of quadratic forms in coordinates ,
we use the language of lattices . we denote in this section by @xmath8 the additive group of the integers modulo @xmath9 and by @xmath10 the ring of @xmath11-adic integers .
recall that a map @xmath12 from an abelian group @xmath13 into an additively written group @xmath14 is called a quadratic form if the expression @xmath15 is additive in @xmath16 and in @xmath17 , i.e. , bilinear . a _ finite quadratic space _
@xmath18 is a finite abelian group @xmath19 together with a quadratic form @xmath20 , such that the induced bilinear form @xmath21 , @xmath22 is nondegenerate . note that different quadratic forms can induce the same bilinear form .
a finite quadratic space can be decomposed uniquely as a direct sum : @xmath23 where the @xmath24 are the @xmath11-sylow subgroups of @xmath19 and @xmath25 is the quadratic form @xmath12 restricted to @xmath24 . furthermore , each @xmath26 is the direct sum of quadratic spaces as follows : for primes @xmath27 , it is the sum of spaces @xmath28 , with @xmath29 , @xmath30 and quadratic form @xmath31 , @xmath32 , where @xmath33 is an integer with jacobi symbol @xmath34 .
the only relations in the abelian semigroup generated by the @xmath28 for fixed @xmath11 arise from @xmath35 , with @xmath36 , @xmath37 . for @xmath38 , one takes @xmath39 , there are also two other kind of generators and further relations ; cf .
@xcite , prop .
1.8.1 and 1.8.2 .
* definition 2.b * an even lattice is a free @xmath40-module of finite rank together with a nondegenerate symmetric bilinear map @xmath41 , such that @xmath42 for @xmath43 .
the map @xmath44 can be linearly extended to @xmath45 for any ring @xmath46 .
every lattice @xmath47 defines a quadratic space @xmath18 by letting @xmath48 , where @xmath49 is the dual lattice , and @xmath50 is the quadratic form @xmath51 , called the discriminant form . the _ signature @xmath52 _ of @xmath47 is the pair @xmath53 , where @xmath54 respectively @xmath55 are the maximal dimension of a positive respectively negative definite subspace of @xmath56 . by sylvester s law of inertia , @xmath57 , the rank of @xmath47 .
the _ genus of @xmath47 _ is the collection of the local lattices @xmath58 , @xmath11 a prime number , including @xmath59 for @xmath60 .
it follows easily that the genus of @xmath47 determines @xmath61 and @xmath52 . also , the converse is true : the discriminant form @xmath61 and the signature @xmath52 determine the genus of @xmath47 .
not every pair consisting of a finite quadratic space @xmath18 and a pair @xmath53 of nonnegative integers can be realized as the genus of a lattice .
first , there is a condition modulo @xmath62 : for a lattice @xmath47 with discriminant form @xmath61 and signature @xmath53 , one has @xmath63 the next result shows that for ranks large enough , this is the only condition that the quadratic space and the signature have to satisfy .
an even lattice @xmath47 with discriminant form @xmath18 and signature @xmath53 exists if condition ( [ milgram ] ) is satisfied and @xmath64 .
note that @xmath65 is clearly a necessary condition . for @xmath66 ,
one also has a precise but more complicated condition for the existence of a lattice ; see @xcite , th .
1.10.1 .
the set @xmath67 is finite .
the proof uses the result that there are only finitely many lattices @xmath47 of fixed discriminant @xmath68 ; see for example @xcite , ch . 9 , th . 1.1 .
much more precise information about the number of lattices in a genus is known .
we restrict ourselves to positive definite lattices , i.e. , lattices @xmath47 of rank @xmath69 where @xmath70 and the @xmath11-component @xmath71 .
this result was obtained independently by h. j. s. smith @xcite and h. minkowski @xcite . the number of embeddings of lattices @xmath72 of rank @xmath73 into @xmath47 for all @xmath72 can be encoded in the siegel theta series @xmath74 of genus @xmath75 , a function on the siegel upper half plane @xmath76 of genus @xmath75 : @xmath77 it is a modular form of weight @xmath78 for a congruence subgroup @xmath79 of @xmath80 acting on @xmath81 .
@xmath82 where @xmath83 is an explicitly given _
eisenstein series _ for @xmath79 , depending only on the genus of @xmath47 .
this theorem was first proven by c. l. siegel @xcite and reformulated in an adelic picture by a. weil , cf .
@xcite , see also @xcite for a general setting .
there is a slightly finer invariant for lattices , namely m. eichler s _ spinor genus _ , cf .
the set of lattices in a genus can be decomposed into @xmath84 spinor genera , also very often @xmath85 .
the sum @xmath86 over all lattices @xmath87 in the same spinor genus as a positive definite lattice @xmath47 is for all spinor genera in the genus of @xmath47 the same if @xmath88 . for indefinite lattices of rank at least @xmath89 ,
there is only one lattice in each spinor genus .
this is one way to see that there is only one even unimodular ( i.e. , of discriminant @xmath90 ) lattice @xmath91 of signature @xmath92 if @xmath93 .
furthermore , one can deduce that two lattices @xmath47 and @xmath87 are in the same genus if and only if @xmath94 and @xmath95 are isomorphic .
there is also the concept of _ rational equivalence_. two lattices @xmath47 and @xmath87 are called rational equivalent if the rational quadratic spaces @xmath96 and @xmath97 are isomorphic .
an equivalent more geometric formulation is to say that both @xmath47 and @xmath87 have an isomorphic sublattice @xmath72 of finite index . from the hasse principle
, it follows that @xmath47 and @xmath87 are rational equivalent if and only if all the local quadratic spaces @xmath98 and @xmath99 , for @xmath11 a prime including @xmath60 , are isomorphic . in particular , lattices in the same genus are rational equivalent .
finally , we remark that the results explained in this section hold similarly for non - even integral lattices .
in this section , we explain how the concepts from the arithmetic of quadratic forms as explained in the last section can , at least partially , be generalized to vertex operator algebras . a good starting point for the description of three - dimensional topological quantum field theories
are modular braided tensor categories @xcite . with the help of the kirby calculus @xcite
, one can use them to define invariants of @xmath89-manifolds , cf .
we will give the basic definitions , for details see @xcite .
a _ monoidal _ category is a category @xmath100 together with a functorial associative tensor product @xmath101 and a neutral object @xmath102 for @xmath103 .
we assume that the tensor category is _ strict _ , so for all objects @xmath104 , @xmath105 , @xmath106 of @xmath107 the products @xmath108 and @xmath109 are identical and not just isomorphic objects . a _ ribbon category _ is a monoidal category together with : 1 .
functorial isomorphisms @xmath110 ( the _ braiding _ ) such that @xmath111 2 .
functorial isomorphisms @xmath112 ( the _ twist _ ) such that @xmath113 3 . a triple @xmath114 , which associates to any @xmath105 a _ dual object _ @xmath115 and morphisms @xmath116 ( the _ evaluation _ ) and @xmath117 ( the _ coevaluation _ ) , such that @xmath118 @xmath119 these axioms can be visualized by labeled bands in @xmath120 . in a ribbon category
, one can define for any endomorphism @xmath121 the _ trace _ of @xmath122 as an element of @xmath123 : @xmath124 the _ quantum dimension _
@xmath125 of @xmath105 is the trace of the identity morphism of @xmath105 .
a _ modular braided tensor category _
( over @xmath126 ) , or
_ modular category _ for short , is a ribbon category which has also the structure of an abelian category over @xmath126 such that : 1 .
the tensor product is @xmath126-linear and @xmath127 ; 2 .
every object is a direct sum of a finite set of simple objects ; 3 .
the isomorphism classes of simple objects form a finite set @xmath128 and an object @xmath105 is simple precisely if @xmath129 ; 4 .
the matrix @xmath130 is invertible .
we call the number @xmath131 the _ discriminant_. in @xcite , a square root @xmath132 is called a rank .
the @xmath133-matrix is defined as the matrix @xmath134 .
we also need @xmath135 . here , the numbers @xmath136 are defined by @xmath137 .
the @xmath138-matrix is the diagonal matrix @xmath139 with the @xmath140 , @xmath141 , on the diagonal .
the @xmath133- and @xmath138-matrix satisfy the relations @xmath142 , @xmath143 and @xmath144 .
these are the same relations as for the generators @xmath145 and @xmath146 of @xmath147 , the mapping class group of a genus @xmath90 surface .
therefore the @xmath133- and @xmath138-matrix define a complex representation @xmath148 of @xmath147 of dimension @xmath149 .
if we omit the factor @xmath150 in the definition of the @xmath138-matrix , we obtain a more natural projective representation of @xmath147 , cf .
@xcite , ii.3.9 . in the context of modular categories
, one can prove that the @xmath151 and @xmath152 are roots of unity ; cf .
@xcite . finally , there is the notation of a _ unitary modular braided tensor category _ where one has natural maps @xmath153 identical to complex conjugation for @xmath154 , @xmath141 , such that @xmath155 for all @xmath156 .
two modular categories are called equivalent , if there is a functorial isomorphism between them , carrying over all structures .
one can also consider nonstrict monoidal categories and define modular categories by using them .
every monoidal category is equivalent to a strict one by maclane s coherence theorem and this is also true for modular braided tensor categories , cf .
@xcite , xi.1.4 .
the categories arising from conformal field theory are in general not strict .
we formulate two versions of the _ verlinde formula _ which are theorems in the context of modular categories .
first , let @xmath157 be the structure constants of the _ fusion algebra _ defined on the free vector space @xmath158 $ ] by the product @xmath159 then one has @xmath160 where @xmath161 is the label defined by @xmath162 and @xmath163 is the label for @xmath164 .
secondly , let @xmath165 be the dimension of the vector space associated by the @xmath166-tqft to a genus @xmath75 surface decorated with labels @xmath167 , @xmath168 , @xmath169
. then one has @xmath170 we mention three important classes of examples of modular categories : \1 ) the modular category @xmath171 associated to a finite quadratic space @xmath18 , see @xcite , i.1.7.2 and ii.1.7.2 .
we have @xmath172 , @xmath173 , where the maps @xmath174 , @xmath175 ( @xmath176 denotes the @xmath177-torsion subgroup of @xmath19 ) used in @xcite are up to equivalence identical to the map @xmath12 , see @xcite , sec . 7.5 .
since the fusion algebra of @xmath171 is just the group ring @xmath178 $ ] of the abelian group @xmath19 , we call such a modular category abelian . any modular category with such an abelian group ring
@xmath178 $ ] as fusion algebra arises in such a way , see @xcite , ch . 7 .
\2 ) the chern - simons theory associated to a simple lie group @xmath14 and positive integral level @xmath179 .
this modular category can be constructed using quantum groups , cf .
@xcite .
\3 ) discrete chern - simons theories constructed from a finite group @xmath14 and a class @xcite .
find a good description of the set of equivalence classes of modular braided tensor categories over @xmath180 .
for two such categories @xmath107 and @xmath181 , it is easy to define a product category @xmath182 , such that its simple objects are the products of the simple objects of the factors .
one can decompose a category into indecomposable pieces @xmath183 under the product : @xmath184 .
an additional problem compared to the situation for quadratic spaces is that it is not possible to decompose @xmath107 into a product of local pieces @xmath185 for different primes @xmath11 such that , for example , the numbers @xmath151 for @xmath185 are @xmath186-th root of unity .
a conjecture for modular braided tensor categories ( also usually formulated for rational voas ) says that the above introduced representation @xmath187 has a congruence subgroup as kernel or , more precisely , @xmath188 where @xmath9 is the smallest natural number such that @xmath189 .
away from the prime @xmath38 , this has been proven by coste and gannon @xcite .
bantay @xcite has recently reduced the problem to a property of permutation orbifolds . assuming this conjecture
, eholzer @xcite has classified the indecomposable modular categories ( or at least the fusion algebras together with the @xmath138-matrix and @xmath152 ) with @xmath190 using the representation theory of @xmath191 .
a lot of work has been done on the classification of fusion algebras and also compatible tensor categories , cf .
for example @xcite . for the purpose of this paper
, the following definition of a restricted class of vertex algebras seems to be most adequate .
a _ vertex operator algebra of central charge @xmath192 _ is a graded complex vector space with @xmath193 together with a linear map @xmath194 $ ] ( _ the state field correspondence _ )
such that for @xmath195 the coefficients @xmath196 of @xmath197 are of degree @xmath198 and @xmath199 for fixed @xmath200 and @xmath201 large enough , and there are two distinguished elements @xmath202 ( the _ vacuum _ ) and @xmath203 ( the _ virasoro element _ ) subject to the following axioms : * @xmath204 and for @xmath205 , one has @xmath206 for @xmath207 , @xmath208 . * for @xmath209 , @xmath205
, there is a @xmath210 with @xmath211 in @xmath212 $ ] . *
the coefficients of @xmath213 define a representation of the _ virasoro algebra _ of central charge @xmath192 : @xmath214=l_{m+n}+\frac{m^3-m}{12}\delta_{m+n,0}\cdot c\cdot { \rm id}_v$ ] and for @xmath205 , one has @xmath215 .
these axioms can be distilled from the general axioms of axiomatic quantum field theory applied to the case of riemannian surfaces as the space time . to obtain a more geometric formulation as indicated in the introduction from a voa
, further conditions have to be satisfied . for the terms we use below and for details
, we refer to the literature : there is the notion of modules and intertwining spaces @xmath216 for a voa , see @xcite . a voa is usually called _ rational _ , if there are only finitely many nonisomorphic irreducible modules and any module can be decomposed into a finite direct sum of irreducible ones .
( the first condition can be deduced from the second , cf .
@xcite . ) the _ conformal weight
_ @xmath217 of a module is its smallest @xmath218-eigenvalue .
let @xmath219 be a complete set of nonisomorphic irreducible modules .
the @xmath138-matrix is defined as the diagonal matrix @xmath220 with the numbers @xmath221 , @xmath222 , on the diagonal .
zhu has shown @xcite that under some conditions on the rational voa , there is a representation of @xmath223 on some space of genus @xmath90 correlation functions .
the generator @xmath224 of @xmath223 defines the matrix @xmath225 . using the tensor product theory for modules of voas ,
huang and lepowsky associated a ( nonstrict ) braided tensor category to a suitable class of voas , see @xcite and the references in @xcite .
finally , there is a definition for the dimension @xmath226 of the space of vacua on the genus @xmath75 surface with labels @xmath167 , @xmath168 , @xmath169 by zhu @xcite .
i assume further that the voa is _ unitary _ , this means the voa can be defined over the real numbers and the natural invariant symmetric form on it is positive definite , although this will exclude many interesting examples .
also , the voa should be _ simple _ , i.e. , @xmath105 is irreducible as a module over itself .
we like to use the above mentioned results and some other properties believed to be true for many interesting voas .
we therefore make the following assumption on the voas considered : [ hauptvermutung ] the voa @xmath105 satisfies the conditions needed in the construction of the braided tensor category on the category of modules as in @xcite and also satisfies the conditions used in @xcite . furthermore , the braided tensor category of @xmath105-modules has the structure of a modular braided tensor category @xmath227 such that @xmath228 one problem is certainly to find simple sufficient conditions on the voa so that the assumption becomes a theorem .
sometimes other conditions in the definition of rationality are added .
assuming that assumption [ hauptvermutung ] is true would directly imply that the central charge and conformal weights are rational numbers .
property ( iv ) follows from ( iii ) if one can prove the sewing relations @xmath229 and @xmath230 for important examples of voas , at least parts of the assumption are known to be true ; for details we refer to the references : \1 ) voas defining an abelian intertwining algebra ; cf .
@xcite .
\2 ) wess - zumino - witten models , i.e. , the voas defined on the highest weight representations of level @xmath179 for an affine kac - moody lie algebra @xmath231 ; cf . @xcite .
\3 ) holomorphic orbifolds ; cf .
@xcite .
\4 ) minimal models ; cf . @xcite .
one could try to generalize the above mentioned results to rational vertex algebras . in their definition one
allows @xmath40-graded vector spaces @xmath105 and the pieces @xmath232 are not assumed to be finite - dimensional .
vertex algebras should be considered as the analog of indefinite even lattices . for a vertex algebra @xmath105 of central charge @xmath192
, one can define the number @xmath233 as the largest central charge of a subvoa @xmath106 of @xmath105 , @xmath234 as @xmath235 and call the pair @xmath236 the signature of @xmath105 . for the rest of this section
, we consider only voas for which assumption [ hauptvermutung ] is satisfied .
the _ genus _ of a rational voa @xmath105 of central charge @xmath192 is defined as the pair consisting of the associated modular braided tensor category @xmath227 and the central charge @xmath192 .
it is not known much about the set of central charges which can occur . from the discussion of modular categories , we know that @xmath192 has to be rational .
the theory of unitary highest weight representations of the virasoro algebra shows that @xmath192 belongs to the minimal series @xmath237 , @xmath238 , @xmath239 , @xmath168 , for @xmath240 .
there is the following analogue of milgrams theorem for modular braided tensor categories .
[ thvoamilgram ] for a rational voa @xmath105 with associated modular category @xmath227 and central charge @xmath192 , one has @xmath241 * proof : * the result is an immediate consequence of assumption [ hauptvermutung ] and the existence of the @xmath223-representations .
we have @xmath242 and @xmath243 because of the @xmath147-representations . using properties ( i ) and ( ii ) in assumption [ hauptvermutung ] , we get @xmath244 .
let @xmath107 be a modular braided tensor category .
does there exist rational voas @xmath105 of central charge @xmath192 with @xmath245 if condition ( [ voamilgram ] ) is satisfied and @xmath192 is large enough ? what are the possible values of @xmath192 for abelian @xmath246 ?
for voas there are more central charges @xmath192 possible than one can realize by lattice voas .
the bound @xmath247 does not hold in general : the fixpoint subvoa @xmath248 of the lattice voa @xmath249 for the involution @xmath250 lifted from the reflection at the origin of @xmath251 ( cf .
@xcite ) has central charge @xmath62 , but @xmath252 , where @xmath12 is isomorphic to @xmath253 copies of the hyperbolic plane .
the set @xmath254 is finite .
_ example : _ let @xmath107 be the trivial modular category , i.e. , there is only one simple object .
the corresponding voas are called self - dual or holomorphic . by theorem [ thvoamilgram ] ,
the central charge @xmath192 has to be divisible by @xmath62 .
an interesting case is @xmath255 .
schellekens has found a list of @xmath256 candidates for the corresponding genus @xcite . for the @xmath257 lattice voas constructed from the genus of the positive definite even unimodular lattices in dimension @xmath257 ,
one has an existence and uniqueness theorem .
@xmath258 other voas are completely constructed as @xmath259-orbifolds , cf .
one of them is the moonshine module @xmath260 , cf .
the fact that @xmath260 belongs to the genus was proven in @xcite .
the uniqueness conjecture @xcite that @xmath260 is the only voa @xmath105 in the genus with @xmath261 is still open . for the voa with @xmath262 as the affine kac - moody subvoa , we can show uniqueness , cf .
@xcite and next section .
conformal field theory suggests that for @xmath263 it may be possible to define a ( possibly vector valued ) genus @xmath75 partition function @xmath264 on the genus @xmath75 teichmller space @xmath265 with automorphic properties for the mapping class group of a genus @xmath75 surface . for @xmath266
this follows from zhu s work @xcite .
are there good mass formulas for the genus of a voa of the form @xmath267 one problem is to find a good definition for @xmath268 if @xmath269 is infinite ; one could try to take the tamagawa measure if @xmath269 is a reductive lie group defined over @xmath270 by using voas defined over @xmath271 . for a kind of relative mass formula of this type involving a nonabelian extension problem , see theorem [ relmass ] in the next section .
the definition of the genus of a voa was modeled along the notation of the genus of an even lattice @xmath47 such that the following diagram is commutative @xmath272 the upper horizontal arrow is the lattice construction of voas @xcite , the left down arrow gives the genus of a lattice via the discriminant form , the right down arrow comes from the abelian intertwining algebra structure on the modules of @xmath273 @xcite and the lower horizontal arrow is the map associating the abelian modular category to a finite quadratic space as explained as an example above . regarding genera of vertex algebras , a structure theory seems to be more complicated than for indefinite lattices : the vertex algebras @xmath274 and @xmath275 ( here , @xmath276 denotes the leech lattice ) are not isomorphic since both define in a uniform way different generalized kac - moody lie algebras , namely the monster @xcite and the fake monster lie algebra @xcite .
( they are different because all cartan subalgebras of a generalized kac - moody lie algebra are conjugate by inner automorphisms as shown by u. ray @xcite . ) but both should be self - dual vertex algebras of signature @xmath277 .
another such vertex algebra is studied in @xcite .
a natural generalization of the notion of _ rational equivalence _ from lattices to voas would be as follows : we take the smallest equivalence relation such that two rational voas @xmath105 and @xmath106 of central charge @xmath192 are equivalent if both have isomorphic rational subvoas with the same virasoro element .
this means that there exist voas @xmath278 , @xmath279 , @xmath168 , @xmath280 and voas @xmath281 , @xmath168 , @xmath282 such that @xmath283 is a common subvoa of @xmath284 and @xmath285 for @xmath286 , @xmath168 , @xmath287 .
just to say there exists a common isomorphic subvoa would nt give an equivalence relation : take for @xmath105 and @xmath106 the fixpoint voas @xmath288 and @xmath289 where @xmath104 is the @xmath290 lattice voa and @xmath14 and @xmath291 are finite subgroups of @xmath292 which ca nt be conjugated into the same larger finite subgroup of @xmath292 ( like the binary icosahedral and the binary tetrahedral group ) .
also , one can generalize the discussion of this section to vertex operator super algebras ( svoas ) .
the central charge @xmath192 of a self - dual svoa has to be half - integral @xcite , th .
for an interesting example with @xmath293 and the classification for small @xmath192 see also @xcite .
in this section , we explain how extension problems for suitable voas can be formulated in terms of the associated modular braided tensor category and sometimes solved by this methods . *
the even overlattices @xmath72 of an even lattice @xmath47 are in one - to - one correspondence to the isotropic subspaces @xmath294 of the associated quadratic space @xmath61 . * the quadratic space for @xmath72 is given by the pair @xmath295 . *
two such extensions give isomorphic lattices @xmath72 exactly if the corresponding subspaces can be transformed into each other by an isometry of @xmath61 induced from an automorphism of @xmath47 .
our aim is to present similar results for extensions of voas .
let @xmath105 be a rational voa .
we call a voa @xmath296 an _ extension _ of @xmath105 if it contains a subvoa isomorphic to @xmath105 and has the same vacuum and virasoro element as @xmath105
. we have @xmath297 for a decomposition of @xmath106 into irreducible @xmath105-modules
@xmath298 , where @xmath299 is a complete set of irreducible modules of @xmath105 and the nonnegative integers @xmath300 are the multiplicities of @xmath298 in @xmath106 .
also , we assume that we have fixed an embedding @xmath301 .
we call two extensions @xmath302 and @xmath303 isomorphic if there is an voa - isomorphism @xmath304 and an automorphism @xmath250 of @xmath105 such that @xmath305 . for @xmath306 and @xmath307
, we can define in this way the _ automorphism group _ @xmath308 of the extension @xmath302 .
part ( a ) of theorem [ latticeext ] implies that the overlattices of a lattice can be determined from the structure of its discriminant form .
for voas we have as an analogous result : [ voaext ] the voa - extensions @xmath106 of a rational voa @xmath105 satisfying assumption [ hauptvermutung ] can be determined completely in terms of the associated modular braided tensor category @xmath227 .
* proof : * fix a @xmath105-module isomorphism @xmath309 extending @xmath310 .
the vertex operator @xmath311 is an element of @xmath312 for @xmath296 to be a voa - extension of @xmath105 , the modules @xmath298 with @xmath313 have to be of integral conformal weight and @xmath311 has to satisfy the commutativity axiom .
both conditions are completely controlled by the twist and the braiding of the associated modular braided tensor category .
@xmath314 similarly , one can see that the structure of the intertwining algebra of an extension is completely determined by @xmath227 and @xmath311 .
( see th .
6.2 of @xcite ) using notation from @xcite : ( a ) the extensions @xmath106 of @xmath105 are in one - to - one correspondence to rigid @xmath227-algebras @xmath19 on which the twist @xmath315 is trivial .
( b ) the category of representations of @xmath106 can be identified with @xmath316 .
] we expect that any rational voa @xmath105 has only finitely many nonisomorphic extensions @xmath296 . in the case that @xmath227 is abelian , i.e. , @xmath317 for a finite quadratic space @xmath18 , one could expect in generalization of theorem [ latticeext ] ( b ) that the voa - extensions are determined by the isotropic subspaces of @xmath19 .
this is indeed the case .
recall that a simple module @xmath298 is called a _
simple current _ if for each simple module @xmath318 there is a another simple module @xmath319 such that @xmath320 holds in the fusion algebra .
[ simplecurrent ] let @xmath105 be a rational voa which has an abelian intertwining operator algebra structure on the direct sum of the simple currents .
let @xmath294 be a subgroup of the abelian group @xmath321 of labels of the simple currents for which the modules @xmath322 , @xmath323 , have integral conformal weight .
then there exists a unique simple voa - extension @xmath296 of the form @xmath324 , @xmath325 , and one has @xmath326 .
* proof : * _ existence : _ choose a non - zero vector @xmath327 for and define @xmath328 .
recall @xcite that the first cohomology groups @xmath329 of the second eilenberg - maclane space @xmath330 ( which is up to homotopy the unique cw - complex with homotopy groups @xmath331 , and @xmath332 for @xmath333 ) can explicitly described by the following cochain complex : @xmath334 the coboundary maps @xmath335 which we need are defined by @xmath336 and @xmath337 , where @xmath338 denotes the usual group cohomology coboundary operator .
note that we can work with normalized cochains , i.e. , cochains with @xmath339 , @xmath340 , respectively @xmath341 and @xmath342 for @xmath192 , @xmath343 .
the commutativity and associativity properties for abelian intertwining algebras
@xcite associate to @xmath311 a cocycle @xmath344 .
the operator @xmath311 defines a voa structure on @xmath106 if @xmath345 . the cohomology group @xmath346
can be identified via @xmath347 with the space of quadratic forms @xmath348 on @xmath294 , see @xcite , th .
3 . in our situation , we have @xmath349 since the modules @xmath322 , @xmath323 , have integral conformal weight and @xmath350 equals the exponential function applied to @xmath351 times the conformal weight of @xmath322 .
therefore , there exists a @xmath352 with @xmath353 . replacing @xmath354 by @xmath355 ,
we see that @xmath296 is a voa .
_ uniqueness : _ assume @xmath296 is an extension .
similar as in the proof of prop .
2.5 ( 4 ) of @xcite , one gets @xmath356 for all @xmath192 , @xmath343 . but
@xmath357 since @xmath358 must have multiplicity one .
one has @xmath359 since by proposition 11.9 of @xcite one has @xmath360 if @xmath209 and @xmath361 are not @xmath362 .
we can replace @xmath354 by @xmath363 where @xmath364 is a @xmath89-cocycle , i.e. , @xmath365 .
a @xmath105-module isomorphism of the space @xmath366 defines for any irreducible @xmath322 , @xmath323 , a scalar @xmath367 , i.e. , a two cochain @xmath368 , and @xmath122 is changed to @xmath369 .
but from the universal coefficient theorem , @xmath370 and @xmath371 since the @xmath40-module @xmath372 is divisible .
so , voas @xmath296 for different choices of cocycles @xmath122 are isomorphic .
_ remarks : _ in terms of the associated modular braided tensor category , the simple currents define a subcategory @xmath373 which is abelian , i.e. , @xmath374 for a finite quadratic space @xmath18 ( we allow here that @xmath12 is degenerate , i.e. , @xmath171 has not to be modular ) : the space of quadratic forms on @xmath19 can be identified as before with the cohomology group @xmath375 and one has @xmath376 .
furthermore , @xmath377 since the modules @xmath322 , @xmath323 , have integral conformal weight , i.e , @xmath294 is an isotropic subspace of @xmath18 .
the theorem completely solves the extension problem for voas @xmath105 for which on the direct sum of a complete set of nonisomorphic irreducible modules there is the structure of an abelian intertwining algebra . in this case , the modular braided tensor category @xmath378 of an extension @xmath106 are described as in theorem [ latticeext ] for lattices .
the structure of the modular braided tensor category @xmath378 for general simple current extensions was investigated in @xcite , where a conjecture for the set of irreducible @xmath106-modules and the corresponding @xmath133-matrix was made .
this conjecture was proven in @xcite .
there is a natural action of the dual group @xmath379 on @xmath106 .
let @xmath106 be a voa and @xmath380 be a finite subgroup of the automorphism group .
let @xmath381 be the fixpoint voa .
then the extensions @xmath104 of @xmath105 which are contained in @xmath106 are in one - to - one correspondence to the subgroups of @xmath14 , i.e. , there is a subgroup @xmath382 such that @xmath383 . from the perspective of modular
braided tensor categories , this situation was analyzed independently by bruguires @xcite and mger @xcite .
both give a purely topological characterization of the above situation using a characterization result for neutral tannakian categories obtained by doplicher and roberts @xcite and deligne @xcite , respectively .
both authors describe how to construct the modular braided tensor category @xmath378 from @xmath227 and the representation subcategory for the group @xmath14 .
mger also proves the galois correspondence , see @xcite , sec .
4.2 . as a final example
, we completely solve the extension problem for self - dual framed vertex operator algebras ( fvoas ) .
fvoas were introduced in @xcite and are voas which have a tensor product @xmath384 of @xmath385 central charge @xmath386 rational virasoro voas as subvoa with the same virasoro element , i.e. , they are the extensions of @xmath384 . recall that @xmath387 has three irreducible modules @xmath387 , @xmath388 and @xmath389 of conformal weight @xmath362 , @xmath390 and @xmath391 and @xmath388 is a simple current . from the decomposition @xmath392 one gets the two codes @xmath393 and it was shown in @xcite that this codes are linear , @xmath394 ( @xmath395 denotes the code orthogonal to @xmath294 ) , @xmath107 is even , and all weights of vectors in @xmath181 are divisible by @xmath62 .
theorem [ simplecurrent ] says that there is a unique fvoa of the form @xmath396 for every linear even code @xmath107 , a result proven in @xcite ( prop . 2.16 ) and
also obtained in @xcite .
as mentioned above , the associated modular category for @xmath397 is described in @xcite .
the set of irreducible modules is described in prop .
5.2 there , a result first proven in @xcite . in special cases ,
the fusion algebra and the @xmath133-matrix and the intertwining algebra were determined in @xcite .
[ relmass ] for @xmath398 , one has @xmath401 where the sum on the left hand side of the equation runs over all isomorphism classes ( in the sense of extensions ) @xmath105 of self - dual fvoas of central charge @xmath402 and @xmath403 is the number of codes @xmath181 of length @xmath385 with @xmath404 , @xmath405 and * sketch of proof : * since @xmath397 is a simple current extension , its modular category @xmath406 can be expressed completely in terms of the known modular category @xmath407 and the code @xmath107 . for
@xmath105 self - dual , all quantum dimensions of @xmath406 are @xmath90 , so it is an abelian modular category @xmath408 with an explicitly given abelian group @xmath19 and quadratic form @xmath12 .
one has a short exact sequence @xmath409 and the self - dual extensions of @xmath397 are in one - to - one correspondence with sections @xmath410 for which @xmath411 is isotropic . counting them gives the result .
let us check the theorem for the first case @xmath412 . in @xcite , the five fvoas of central charge @xmath62 together with their automorphism group have been determined . for every @xmath413 , @xmath177 , @xmath89 , @xmath239 , @xmath253
, there is an up to equivalence unique code @xmath414 of dimension @xmath179 and a unique fvoa @xmath415 with code @xmath416 and automorphism group @xmath417 . by @xcite ,
main theorem 2 , one has @xmath418 and for the sum @xmath419 we obtain : @xmath420 in agreement with theorem [ relmass ] .
the methods of the proof can be used to give a construction of the moonshine module @xmath260 as the self - dual fvoa with @xmath421 the lexicographic code of length @xmath422 and minimal weight @xmath239 ; see @xcite , sect . 5 for the precise structure of the virasoro module decomposition .
the only input from voa - theory which one needs is the construction of the virasoro voa of central charge @xmath386 and the structure of its intertwining algebra .
all previous constructions use at some place the lattice vertex operator algebra construction ; cf .
@xcite .
igor b. frenkel , yi - zhi huang , and james lepowsky , _ on axiomatic approaches to vertex operator algebras and modules _ , memoirs of the ams , band 104 , nr . 494 , american mathmeatical society , providence , 1993 .
robert griess and gerald hhn , _ virasoro frames and their stabilizers for the @xmath424 lattice type vertex operator algebra _ , to appear in journal fr die reine und angewandte mathematik , math.qa/0101054 ( 2001 ) .
yi - zhi huang , _ a nonmeromorphic extension of the moonshine module vertex operator algebra _ ,
moonshine , the monster , and related topics ( south hadley , ma , 1994 ) , amer .
soc . , providence , ri , 1996 , hep - th/9406190 , pp .
123148 .
saunders maclane , _ cohomology theory of abelian groups _ , proceedings of the international congress of mathematicians , cambridge , mass . , 1950 ,
vol . 2 ( providence , r. i. ) , amer .
soc . , 1952 , pp .
814 .
akihiro tsuchiya , kenji ueno , and yasuhiko yamada , _ conformal field theory on universal family of stable curves with gauge symmetries _ , integrable systems in quantum field theory and statistical mechanics , adv .
pure math .
, 19 , academic press , boston , ma , 1989 , pp .
459566 .
yongchang zhu , _ vertex operator algebras , elliptic functions , and modular forms _ ,
thesis , yale university , 1990 , appeared as : _ modular invariance of characters of vertex operator algebras , _
soc * 9 * ( 1996 ) . | the notion of the _ genus _ of a quadratic form is generalized to vertex operator algebras .
we define it as the modular braided tensor category associated to a suitable vertex operator algebra together with the central charge .
statements similar as known for quadratic forms are formulated .
we further explain how extension problems for vertex operator algebras can be described in terms of the associated modular braided tensor category . |
we thank richard friend , ayelet vilan and dassia egorova , as well as the reviewers , for useful discussions , and johannes zimmermann and tobias glaser for the angle - dependent ir spectra .
this work was supported by the netherlands organization for scientific research onderzoek ( nwo ) through the `` stichting voor fundamenteel onderzoek der materie '' ( fom ) research program .
also acknowledges a veni grant from the nwo .
is currently a royal society university research fellow .
r.l . acknowledges a marie curie ie fellowship from the eu , held at the weizmann institute .
thanks the council for higher education ( israel ) for a pbc program postdoctoral research fellowship . v.c . and j.l.b .
thank support from the office of naval research and muri center on advanced molecular photovoltaics , award no .
n00014 - 14 - 1 - 0580 .
thanks the israel science foundation centre of excellence program , the grand centre for sensors and security and the schmidt minerva centre for supramolecular architecture for partial support .
d.c . holds the sylvia and rowland schaefer chair in energy research .
50ifxundefined [ 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 10.1021/ar00118a005 [ * * , ( ) ] link:\doibase 10.1126/science.1146556 [ * * , ( ) ] link:\doibase 10.1038/nature09346 [ * * , ( ) ] link:\doibase 10.1371/journal.pone.0055780 [ * * , ( ) ] link:\doibase 10.1126/science.1249771 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.056801 [ * * , ( ) ] link:\doibase 10.1038/nphys2515 [ * * , ( ) ] link:\doibase 10.1038/nature01649 [ * * , ( ) ] link:\doibase 10.1038/ncomms2708 [ * * , ( ) ] link:\doibase 10.1038/nmat3710 [ * * , ( ) ] link:\doibase 10.1038/nnano.2010.240 [ * * , ( ) ] link:\doibase 10.1038/nnano.2012.37 [ * * , ( ) ] link:\doibase 10.1021/jp403005q [ * * , ( ) ] link:\doibase 10.1021/ja1040732 [ * * , ( ) ] link:\doibase 10.1038/ncomms1164 [ * * , ( ) ] link:\doibase 10.1016/j.chemphys.2010.05.021 [ , * * , ( ) ] link:\doibase 10.1038/nature06262 [ * * , ( ) ] link:\doibase 10.1146/annurev.physchem.58.032806.104456 [ * * , ( ) ] link:\doibase 10.1039/c3cp50871d [ * * , ( ) ] link:\doibase 10.1021/ja074657x [ * * , ( ) ] link:\doibase 10.1038/nature06119 [ * * , ( ) ] link:\doibase 10.1038/srep03823 [ * * ( ) , 10.1038/srep03823 ] link:\doibase 10.1038/nature03429 [ * * , ( ) ] link:\doibase 10.1038/nchem.1881 [ * * , ( ) ] link:\doibase 10.1038/nature08811 [ * * , ( ) ] link:\doibase 10.1364/josab.28.000171 [ * * , ( ) ] link:\doibase 10.1126/science.1217745 [ * * , ( ) ] link:\doibase 10.1016/0009 - 2614(81)80471-x [ * * , ( ) ] link:\doibase 10.1103/physrevb.62.2505 [ * * , ( ) ] link:\doibase 10.1103/physrevb.72.195208 [ * * , ( ) ] link:\doibase 10.1364/oe.21.028617 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2013.151 [ * * , ( ) ] link:\doibase 10.1016/j.chemphys.2012.08.019 [ , * * , ( ) ] link:\doibase 10.1021/ar300345h [ * * , ( ) ] link:\doibase 10.1002/adma.201000376 [ * * , ( ) ] link:\doibase 10.1063/1.2403241 [ * * , ( ) ] link:\doibase 10.1002/adma.201403556 [ * * , ( ) ] link:\doibase 10.1126/science.287.5454.839 [ * * , ( ) ] link:\doibase 10.1016/0301 - 0104(95)00290 - 1 [ * * , ( ) ] link:\doibase 10.1021/jp048343x [ * * , ( ) ] link:\doibase 10.1103/physrevb.70.081309 [ * * , ( ) ] link:\doibase 10.1038/990058 [ * * , ( ) ] link:\doibase 10.1103/physrev.120.745 [ * * , ( ) ] link:\doibase 10.1073/pnas.092143399 [ * * , ( ) ] link:\doibase 10.1016/s0379 - 6779(99)00050 - 8 [ * * , ( ) ] link:\doibase 10.1107/s0365110x62000699 [ * * , ( ) ] link:\doibase 10.1016/j.apsusc.2004.10.131 [ , * * , ( ) ] link:\doibase 10.1103/physrevb.75.174108 [ * * , ( ) ] link:\doibase 10.1021/jp900157p [ * * , ( ) ] `` , '' in link:\doibase 10.1002/9781119953678.rad046 [ _ _ ] ( , ) | * a.a.b . , r.l . , and y.x . contributed equally .
+ + the soft character of organic materials leads to strong coupling between molecular nuclear and electronic dynamics .
this coupling opens the way to control charge transport in organic electronic devices by exciting molecular vibrational motions . however , despite encouraging theoretical predictions , experimental realization of such control has remained elusive .
here we demonstrate experimentally that photoconductivity in a model organic optoelectronic device can be controlled by the selective excitation of molecular vibrations . using an ultrafast infrared laser source to create a coherent superposition of vibrational motions in a pentacene / c60 photoresistor
, we observe that excitation of certain modes in the @xmath0{cm^{-1}}$ ] region leads to photocurrent enhancement .
excited vibrations affect predominantly trapped carriers .
the effect depends on the nature of the vibration and its mode - specific character can be well described by the vibrational modulation of intermolecular electronic couplings .
vibrational control thus presents a new tool for studying electron - phonon coupling and charge dynamics in ( bio)molecular materials .
+ * submitted * : nov .
11 , 2014 + * revised * : feb . 3 , 2015 the soft character of organic materials strongly influences their electronic functionality .
@xcite in these systems charge hopping and electronic delocalization are determined by the overlap of the molecular orbitals and , therefore , is highly sensitive to minor changes in molecular geometry .
hence , the electronic properties of organic materials are largely determined by the interplay between the electronic and nuclear dynamics of the molecules , referred to as vibronic coupling phenomena .
a growing number of interdisciplinary studies show that vibronic effects lie at the heart of a diverse class of effects in physics , chemistry and biology - from non - linear behavior of molecular junctions @xcite to photophysics of vision @xcite and even olfactory reception .
@xcite vibrational motions have been postulated to regulate the interaction between different molecular electronic states by modulating inter- and intra - molecular couplings , by donating or accepting extra energy quanta @xcite , and by suppressing @xcite or promoting @xcite quantum interference phenomena .
vibronic effects were also shown to be fundamentally important for the conductivity of organic materials .
vibrational motions influence intermolecular electron tunneling probabilities @xcite and govern a variety of non - equilibrium phenomena such as local heating @xcite , switching @xcite , hysteresis , and electronic decoherence @xcite .
this makes vibrational excitation a promising tool for spectroscopy of molecular junctions @xcite , tracking charge transfer processes in organic- and bio - electronic systems , and , more generally , for the development of electronic devices . for example , remarkable opportunities for organic electronics would arise from the possibility to control charge transport , and , thus , affect device performance by coherently driving nuclear motions along a pre - selected reaction coordinate trajectory .
however , despite many encouraging theoretical predictions @xcite the experimental realization of vibrationally controlled electronics is still elusive due to the complexity of selective control of nuclear motions in an actual electronic junction . until now
, vibration - associated charge dynamics in organic electronic devices have been only controlled with approaches that do not include mode selectivity .
for example , the density and the equilibrium population of vibrational states have been varied via chemical synthesis of molecules with different bond structures @xcite and via thermal population of low - frequency vibrations .
@xcite however , in principle , it should be possible to access particular non - equilibrium nuclear or vibronic states by using instrumentation of optical time - resolved techniques , like visible pump - probe @xcite , time - resolved stimulated / impulsive raman @xcite , or transient ir absorption @xcite .
for example , for inorganic perovskite materials @xcite and molecular mott insulators @xcite it has been reported that selective ir excitation can lead to strong modulation of the electronic properties by inducing a lattice phase transition .
sophisticated all - optical two - dimensional ( 2d ) photon echo techniques are even capable of guiding a molecular system through a desired quantum superposition of vibronic / vibrational states @xcite .
although such spectroscopic methods provide a comprehensive approach for probing and controlling molecular motions and have been applied to model systems such as molecular thin films or solutions , they have not yet been employed to control functional electronic ( nano)devices . in this work ,
we combine device characterization and ultrafast - spectroscopy methods to experimentally demonstrate that the performance of an organic optoelectronic system can be controlled by selectively exciting vibrational modes of the molecules involved in charge transport . as model system we use pentacene / c60 bi - layer photoresistors .
our experimental approach is based on the interferometric extension of the pump - push photocurrent ( ppp ) technique .
@xcite in a ppp experiment , an optoelectronic device is illuminated by a sequence of laser pulses interacting with the active material in the device .
the result of these interactions is detected by observing the variations in the current flow through the device as a function of time delay @xmath1 between the pump and push pulses and their spectra .
thus , ppp combines the sensitivity and device relevance of electronic methods with the excitation selectivity and ultrafast time resolution of optical methods . since its introduction @xcite ,
ppp has been applied and discussed in the context of photovoltaics @xcite , nanoelectronics , spectroscopy @xcite , microscopy @xcite , and molecular junction research @xcite . in this work ,
we extend the ppp method , using the recent progress in ultrafast interferometry @xcite that allows for a precise control over the time / frequency - domain structure of the ir optical pulses .
we apply a sequence of ultrafast mid - ir laser pulses to create a coherent superposition of molecular vibrational motions inside the active layer of a device and correlate this excitation with the device performance .
the molecular electronic device characterisation : ( a ) molecular arrangement of molecules in the pentacene crystal and c60 fullerene structure .
( b - c ) layout and microscope image of the device .
( d ) ir absorption in the vibrational fingerprint region and optical absorption spectra of pentacene and c60 .
the yellow shaded contour shows a typical laser spectrum used for ir push .
( e ) photocurrent from the device as a function of visible light modulation frequency ; the line is a cole - cole fit with a @xmath2{ms}$ ] lifetime constant and @xmath3 dispersion parameter . ]
figure [ fig1 ] a)-c ) describes the organic bilayer photoresistor model system .
the active layer of the device consists of polycrystalline pentacene ( @xmath4{nm}$ ] ) and fullerene c60 ( @xmath5{nm}$ ] ) films ( fig .
[ fig1 ] a ) ) , thermally evaporated on top of 3- , 5- or @xmath6{\upmu m}$ ] spaced electrodes arranged in a comb - like geometry on a sio@xmath7 substrate ( fig .
[ fig1 ] b ) ) .
we chose this geometry rather than a sandwich - like structure , typical for photodiodes or solar cells , to improve the access of mid - ir pump - pulses to the active layer . adding the c60 layer was critical to enhance the photocarrier generation in the film .
@xcite figure [ fig1 ] d ) compares the absorption spectra of pentacene and c60 in the ir vibrational fingerprint region and in the region of the optical electronic transitions .
c60 shows several distinct vibrational modes at @xmath8{cm^{-1}}$ ] , @xmath9{cm^{-1}}$ ] , and @xmath10{cm^{-1}}$ ] and has a comparably low optical density in the visible .
pentacene has a rich spectrum of vibrational lines in the ir and also shows strong excitonic absorption features at frequencies above @xmath11{cm^{-1}}$ ] ( @xmath12{nm}$ ] ) .
according to density functional theory calculations , the strong ir peaks at 1300 and @xmath13{cm^{-1}}$ ] are mostly associated with @xmath14 stretching vibrations along the short axis of pentacene , while the weaker high - frequency vibrations correspond to atomic motions mostly aligned with the long axis of the molecule ( see si ) .
the dark i - v curves of the devices are symmetric and roughly linear , indicating good hole injection from the gold electrodes to the pentacene layer ( see si ) .
upon exposure to visible light , the current flow through the devices strongly increases ( @xmath15 times under @xmath6{mw / cm^2}$ ] illumination ) .
devices without a c60 layer demonstrated only negligible photoconductivity , which indicates that singlet ( and triplet ) @xcite excitons generated after pentacene excitation are dissociating at the pentacene / c60 interface and that the charge generation proceeds through the interfacial charge transfer states @xcite .
due to the large electron injection barrier at the pentacene / au interface , the dark current is mostly provided by holes , while under illumination both holes and electrons contribute to the photocurrent . unlike in a typical solar cell , both electrodes are placed below the pentacene films .
therefore , electrons and holes have to pass through the pentacene , which is known to lead to extremely long ( up to seconds ) extraction times of electrons residing in low - lying trap states in pentacene .
@xcite this notion is confirmed by the dependence of the photocurrent on the light - modulation frequency ( fig .
[ fig1 ] e ) ) .
cole - cole analysis of this dependence shows a typical time constant @xmath16{ms}$ ] , which we interpret as the lifetime of long - lived electronic charge carriers .
figure [ fig2 ] a ) shows the layout of the experiment , designed to observe the effect of molecular vibrations on the charge transport through the device .
the setup combines a @xmath17{khz}$ ] visible - infrared ultrafast spectrometer and a lock - in current probe station wired to the device under @xmath18{v}$ ] external bias .
first , a visible ( @xmath19{cm^{-1}}$ ] ; @xmath20{nm}$ ] ; @xmath21{ev}$ ] ) pump pulse illuminates the device .
the absorption of the pump light in pentacene leads to the build - up of excitons and charge carriers in the active layer .
the generated carriers produce a sequence of @xmath17{ms}$]-spaced current pulses in the measurement circuit with an average photocurrent @xmath22{na}$ ] , detected by the lock - in amplifier at @xmath17{khz}$ ] .
we note that at such low current densities a charge - induced phase transition @xcite can be excluded .
the device is irradiated with a push pulse at certain delay times @xmath1 before or after the pump pulse .
the push pulse can promote @xmath23 of the molecules to the excited vibrational state and can also excite low - frequency charge - associated ir electronic transitions .
@xcite the effect of ir light on the charge separation and transport was detected via the variation of device photocurrent @xmath24 .
figure [ fig2 ] c ) presents a typical ppp transient , measured with a single - pulse push ( one interferometer arm blocked ) at @xmath25{cm^{-1}}$ ] .
when the pump was blocked we observed no signal due to the push only . at negative delay time @xmath1 ,
when the push pulse arrives before the pump , we already observe a substantial increase of the current due to ir excitation ( i.e. , @xmath26 ) .
we associate this response with the excitation of long - lived photocarriers that were generated by the preceding pump pulse that arrives @xmath27{ms}$ ] earlier .
this observation is in line with the long collection times of trapped carriers observed for electrons in pentacene .
@xcite at delay time @xmath28 , the ppp response promptly increases as the concentration of charges in the cell rises due to the arrival of the new pump pulse and the ir push influences their dynamics .
the rapid rise is followed by a @xmath29{ps}$ ] decay component that we assign to the geminate recombination of newly generated charge pairs , which are likely to form electrostatically bound charge - transfer excitons .
@xcite in a broadband experiment using a single - pulse push , it is not possible to distinguish the effects of low - frequency electronic excitations from the vibronic phenomena associated with the interference between the molecular vibrational motions and charge dynamics . to separate and address these phenomena individually ,
we performed push - frequency resolved measurements by exploiting the ultrafast interferometry approach .
@xcite using a mach - zehnder scheme ( fig .
[ fig2 ] a ) ) , the push beam is split into two pulses displaced in time by an interferometric delay @xmath30 .
this leads to the formation of a 1/@xmath30 periodic modulation in the total push spectrum ( fig .
2b ) which allows for selective excitation of different coherent superpositions of modes within the bandwidth of the ir light . in a typical experiment , for a certain pump - push delay
@xmath1 , the signal @xmath31 is detected as a function of interferometric delay @xmath30 ( fig .
[ fig2 ] d ) ) .
the obtained interferogram is fourier - transformed along the @xmath30 axis to yield the action spectrum of the push effect .
figure [ fig2 ] e ) shows a typical frequency - resolved ppp response of a pentacene / c60 device at negative and positive pump - push delay times @xmath1 , and with no pump ( dark ) . at both delays the response consists of a number of narrow peaks on top of a broad featureless response , roughly following the ir source spectrum .
we associate the broad feature with intraband electronic and polaronic absorption , which typically spreads between 1000 and @xmath32{cm^{-1}}$ ] .
@xcite the intraband excitation brings the associated charge carriers to a higher - lying delocalized state , thereby enhancing their mobility , decreasing their recombination , and thus increasing the current output .
@xcite the narrow features in the ppp signal match well with the absorption peaks of the vibrational modes of pentacene and c60 .
therefore these features in the frequency - resolved ppp response are assigned to the excitation of molecular vibrations that modulate the electronic dynamics .
interestingly , the broad electronic response dominates the ppp signal when the push arrives after the pump , while the vibrational features have similar amplitudes ( within the experimental accuracy ) at positive and negative @xmath1 delays .
this observation indicates that the ir electronic excitation substantially promotes charge separation at the pentacene / c60 interface , soon after exciton generation . at the same time , the effect of vibrational excitation is present for long - lived trapped charge carriers and , therefore , does not influence charge separation , but only carrier de - trapping dynamics .
@xcite
we now focus on the analysis of the vibrational features only .
the effect of broadband electronic ir excitation on charge dynamics in organic semiconductors has been investigated previously @xcite , and is outside the scope of this paper . to study the effect of vibrational excitation for a broader set of vibrational modes , we use a wide push spectral window of @xmath33{cm^{-1}}$ ] and long @xmath30-scanning to obtain high frequency resolution .
we also applied time - domain filtering ( see si ) to suppress broad features due to electronic excitation and non - linear field - induced tunneling currents .
experimental evaluation of vibrational control effect : ( a ) the vibrational part of the ppp response , measured at negative delay time ( @xmath34{ps}$ ] ) .
the signal amplitude is normalized to the spectral density of the ir push source .
the spectrum is obtained using two ppp spectra , each covering a different but overlapping wavenumber range ; these are scaled to match the amplitude of the @xmath9{cm^{-1}}$ ] mode that is present in both spectra . for comparison ,
the absorption spectrum of the pentacene / c60 layer is presented in red .
( b ) the influence of different vibrations on device photocurrent , estimated by normalizing the amplitude of the ppp signal to the absorbed ir intensity .
the change of photocurrent absolute value corresponds to a flat @xmath35{j / cm^2}$ ] per @xmath36 spectral density of exciting ir light , fully absorbed by the vibrations .
the error bars are calculated from standard deviations for measurements on different devices ; the number of measurements was 10 for @xmath37{cm^{-1}}$ ] modes and 4 for all other modes . ]
figure [ fig3 ] a ) presents the vibration - associated ppp spectrum covering most of the ir fingerprint frequency range .
the amplitude of the ppp response was normalized to the spectral density of the push pulse to allow for a direct comparison of the different vibrational lines .
the spectrum is the result of several measurements with different push frequencies spliced together to match the amplitude of the effect for the @xmath9{cm^{-1}}$ ] feature , which was present in all measurements . in the @xmath33{cm^{-1}}$ ] region , we observe twelve ppp peaks at frequencies that match well the ir - active vibrational modes of pentacene and c60 .
we note that , while the vibrations of charged pentacene may differ from those of the neutral molecules @xcite , these differences should not be observed in the ppp data .
firstly , the shift in frequency for most individual modes is small @xcite and , for most modes , below our frequency resolution ( @xmath38{cm^{-1}}$ ] ) .
this conclusion is also supported by dft calculations , see suppl .
information , tables s1 and s2 . secondly , the minor shifts of the vibrational levels lead to a highly efficient vibrational energy transfer @xcite between neutral and charged molecules , which allows the vibrational excitation of neutral pentacene to be delivered to the charge trapping sites .
thirdly , according to miller - abrahams formalism , when a carrier hops from a radical to a neutral state all vibrational modes coupled to these electronic states contribute to the transfer rate .
@xcite therefore , it is not surprising that the lattice vibrations , i.e. those of the neutral pentacene , are observed in the detrapping dynamics .
we observe that the amplitude of the ppp response does not follow the intensity of the ir absorption .
for example , the band at @xmath13{cm^{-1}}$ ] possesses a much stronger ir absorption than the @xmath39{cm^{-1}}$ ] vibration , but shows a weaker ppp response .
this result shows that the observed ppp response can not be explained by the equilibration of vibrational energy between modes and average heating of the device active layer , thus illustrating the mode - selective character of the ppp response .
this example illustrates that different atomic motions couple differently to the charge dynamics of the system . to exclude that the non - scaling of the ppp response with ir absorption is merely an effect of a different orientation of the vibration dipoles with respect to the exciting ir light
, we also performed angle - dependent ir absorption measurements .
these measurements showed that the modes exhibiting very different ppp effect , e.g. at @xmath13{cm^{-1}}$ ] and @xmath39{cm^{-1}}$ ] , have similar dipole orientations ( see si ) , which rules out orientation effects .
figure [ fig3 ] b ) compares the effect of vibrational excitation on the device photoconductivity for different vibrational modes , obtained by normalizing the ppp response to the number of photons absorbed by the vibrational mode . in accordance with figure 3a ,
the 1300 and @xmath13{cm^{-1}}$ ] modes show the weakest coupling .
the higher - frequency vibrations of pentacene show a 5 - 8 times higher effect on the photoconductivity . for two of the fullerene vibrations
the effect is similar to that of the high - frequency vibrational modes of pentacene .
these results can be rationalized in the framework of the phonon - assisted miller - abrahams ( ma ) theory .
@xcite according to this model , carrier hopping from a trapping state to higher - energy ( more conducting ) states takes place via absorption of a phonon with energy @xmath40 to compensate for the energy difference between initial and final electronic states ; the hopping rate @xmath41 is defined by the electron - vibrational coupling constant ( @xmath42 ) and the occupation number ( @xmath43 ) of the absorbed phonon , i.e. @xmath44 in thermal equilibrium , the occupation number of a high - energy molecular vibration is very small : @xmath45 for a comprehensive description of the ppp response the interaction with the ir photons should be included into the ma model .
however , at the conceptual level the effect can be understood by assuming that an ir excitation creates a non - equilibrium population of the molecular vibrational manifold ; therefore , an increase in the hopping probability is expected .
the mode - selective character of the ppp response is therefore defined by the electron - vibration coupling constants . to achieve mechanistic insight into the observed phenomena we performed a theoretical analysis of the coupling between the different molecular vibrations and the charge carriers ( holes ) in pentacene . in molecular systems
these couplings can be divided into two types , i.e. local ( holstein - type ) and non - local ( peierls - type ) .
@xcite the holstein electron - phonon interaction originates from the modulation of the site energies by the vibrations .
only totally symmetric molecular vibration modes can contribute to this interaction .
for centrosymmetric molecules like pentacene , the symmetric modes are not ir - active and will not absorb ir photons , so that we can rule out holstein electron - phonon coupling effects in the ppp response .
the peierls - type electron - phonon couplings are associated with the dependence of the transfer integrals on the distances between adjacent molecules and their relative orientations .
@xcite for this type of coupling , there are no symmetry restrictions .
based on previous studies @xcite , we used a triclinic polymorph @xcite to represent the pentacene layer structure in the calculations .
figure [ fig4 ] a ) shows the simulated ir spectra for a single pentacene molecule and for the crystal in comparison to the experimental ir absorption .
the agreement between the experimental and calculated vibrational frequencies ( with typical discrepancies @xmath46{cm^{-1}}$ ] ) allows the assignment of the vibrational modes observed in the frequency - resolved ppp experiment .
the 1300 and @xmath13{cm^{-1}}$ ] features in the experimental spectrum are associated ( fig .
[ fig4 ] c ) and si ) with in - plane ring stretching modes along the short axis of pentacene @xcite , while the ir peaks at @xmath10{cm^{-1}}$ ] and @xmath39{cm^{-1}}$ ] are associated with molecular deformations along the long axis of pentacene ( fig .
[ fig4 ] d ) and si ) .
the non - local hole - vibration couplings are defined as the derivatives of the charge transfer integrals with respect to the vibrational coordinates , @xmath47 , and can be computed numerically . @xcite
both the transfer integrals and electron - vibration couplings have been derived in a one - electron approximation ( see si for details ) .
our results indicate that there are two main transfer integrals contributing to charge transfer in the pentacene crystal : @xmath48{mev}$ ] and @xmath49{mev}$ ] ; both are associated with intermolecular interactions along the herringbone directions ( see the red and blue arrows in fig .
[ fig4 ] b ) .
the other two transfer integrals oriented along the a - axis are substantially smaller and do not demonstrate substantial modulation by ir - active modes .
( see si ) the derived coupling constants of the ir - active modes are shown in figure 4e .
the couplings in the @xmath50{cm^{-1}}$ ] range are about 2 to 5 times larger than in the @xmath51{cm^{-1}}$ ] range . to link the variations in coupling and the probability to de - trap a charge with a vibrational excitation , we estimated the rates of vibration - induced charge hopping . in the case of two pathways ,
the rate is defined within perturbation theory as : @xmath52 where @xmath53 and @xmath54 correspond to quasi - degenerate molecular vibrations of similar frequencies .
figure [ fig4 ] f ) presents these hopping rates together with the experimental observations from figure [ fig3 ] b ) .
the theoretical results are fully consistent with the experimental data . in particular , they capture well the mode - selective character of the phenomena , with the modes below @xmath9{cm^{-1}}$ ] calculated to have a much smaller impact on charge hopping than the higher - frequency ones . based on the calculations ,
the intermolecular electronic couplings and charge transport in pentacene crystals are seen to be most sensitive to stretching deformations along the long molecular axis , while the stretching deformations along the short molecular axis are less important . in conclusion
, we demonstrated that the vibrational coupling phenomena , which play an essential role in molecular - scale charge transport , can be explored and put to action by combining optical and electronic techniques .
both the experiment and theoretical calculations demonstrate that different non - equilibrium geometries and atomic motions have different effects on the charge dynamics .
specifically , our results show that vibrations along the long axis of pentacene molecules lead to a stronger increase of hopping transport via charge de - trapping than vibrations along the short axis .
the mode - selective vibrational control of charge dynamics introduced here opens up a plethora of opportunities for basic research , including the development of high - mobility organic semiconductors , and the utilization of vibronic phenomena for ultrafast switching of organic devices .
in addition , the mode - selective and local nature of our method might be particularly useful for the identification of charge transport mechanisms and pathways in ( bio)molecular junctions . |
it is generally believed , and confirmed by lattice qcd calculations @xcite , that hadronic matter under extreme conditions will form a plasma in which quarks and gluons are no longer confined to individual hadrons and are in both thermal and chemical equilibrium . to search for such a quark - gluon plasma , it is proposed to study collisions of heavy nuclei at extremely high energies .
recent work using models based on perturbative qcd indeed shows that at high energy a dense partonic system can be produced @xcite . though it is not all clear if such partonic systems will reach thermal and chemical equilibrium before hadronization @xcite , the partons , mainly gluons , are certainly in a deconfined state at such high densities @xcite .
many signals could arise from this deconfined state , such as charm quark enhancement @xcite or enhanced photon and dilepton production @xcite .
charm quarks , for example , can not be easily produced during the mixed and hadronic phases of the dense matter , due to their large masses compared to the temperature .
they can be readily produced only during the early stage of the evolution , when partonic degrees of freedom are relevant . in this paper
, we will discuss how pre - equilibrium @xmath0 suppression can be used to probe the early deconfined state of the partonic system and the dynamics governing its evolution toward equilibrium .
@xmath0 suppression due to color screening has been proposed to probe deconfinement @xcite .
this requires that the interactions of @xmath0 with hadrons and deconfined partons are different @xcite .
because of its small size , a heavy quarkonium can probe the short - distance properties of light hadrons .
it is thus possible to make a parton - based calculation of @xmath0-hadron cross section via an operator product expansion method similar to that used in deeply - inelastic lepton - hadron scatterings @xcite .
the resulting @xmath0-hadron cross section can be related to the distribution function of gluons inside a hadron .
the energy dependence of the cross section near the threshold of the break - up of a @xmath0 is determined by the large @xmath2 behavior of the gluon distribution function , giving rise to a very small break - up cross section at low energies . only at very high energies
, this cross section will reach its asymptotic value of a few mb .
in other words , the dissociation can only occur if the gluon from the light hadron wave function is hard in the @xmath0 s rest frame , i.e. its energy is high enough to overcome the binding energy threshold .
a hadron gas with temperature below 0.5 gev certainly can not provide such energetic gluons to break up the @xmath0 .
therefore , a slow @xmath0 is very unlikely to be absorbed inside a hadron gas of reasonable temperature @xcite .
this conclusion does not seem to be affected substantially by non - perturbative effects , analyzed in ref.@xcite .
on the other hand , a deconfined partonic system contains much harder gluons which can easily break up a @xmath0 @xcite .
a study of the energy dependence of the gluon-@xmath0 inelastic cross section @xcite shows a strong peak just above the break - up threshold of the gluon energy , @xmath3 , where @xmath4 and @xmath5 are the @xmath0 and d meson masses , respectively . in the pre - equilibrium stage ,
i.e. , before the partons have reached equilibrium , the average parton transverse momentum is sufficiently large @xcite to break up a @xmath0 , provided the partons are deconfined .
the dissociation of @xmath0 will continue during the whole equilibration process until the effective temperature drops below a certain value or the beginning of hadronization , whichever takes place first
. therefore measurements of @xmath0 suppression can probe the deconfinement of the early partonic system and shed light on the subsequent equilibration process , provided that possible nuclear effects on the production of @xmath6 pairs and on pre - resonance charmonium states are understood and taken into account . in the following we will first calculate the thermal gluon-@xmath0 dissociation cross section at different temperatures and for different @xmath0 transverse momentum .
we then follow the evolution of an initially produced parton gas toward equilibrium and calculate the resulting total survival probability of a @xmath0 and its @xmath7-dependence .
the operator product expansion allows one to express the hadron-@xmath0 inelastic cross section in terms of the convolution of the gluon-@xmath0 dissociation cross section with the gluon distribution inside the hadron @xcite .
the gluon-@xmath0 dissociation cross section is given by @xcite @xmath8 where @xmath9 is the coupling constant of gluon and @xmath10 quark , @xmath11 the @xmath10 quark mass , and @xmath12 the gluon energy in the @xmath13 rest frame ; its value must be larger than the @xmath0 binding energy @xmath14 . since for the tightly bound ground state of quarkonium the binding force between the heavy quark and antiquark
is well approximated by the one - gluon - exchange coulomb potential , the @xmath1 bound state is hydrogen - like and the coulomb relation holds , @xmath15 the cross section thus can be rewritten as @xmath16 as shown in monte carlo simulations @xcite , the parton density in the early stage of high - energy heavy - ion collisions has an approximate bjorken - type @xcite scaling behavior .
we will only consider @xmath0 suppression in the central rapidity region ( @xmath17 ) . in this case , the @xmath13 will move in the transverse direction with a four - velocity @xmath18 where @xmath19 is defined as the @xmath0 s transverse mass .
a gluon with a four - momentum @xmath20 in the rest frame of the parton gas has an energy @xmath21 in the rest frame of the @xmath0 .
the thermal gluon-@xmath0 dissociation cross section is then defined as @xmath22 where the gluon distribution in the rest frame of the parton gas is defined as @xmath23 with @xmath24 specifying the deviation of the system from chemical equilibrium .
the relative velocity @xmath25 between the @xmath0 and a gluon is @xmath26 changing the variable to the gluon momentum , @xmath27 , in the rest frame of the @xmath0 , the integral in the numerator of eq .
( [ eq5 ] ) can be rewritten as @xmath28 where @xmath29 one can carry out the integral in the denominator , @xmath30 , and the angular part in the numerator , to get @xmath31 with @xmath32 and @xmath33 in order to understand the temperature and @xmath7 dependence of the thermal gluon-@xmath0 dissociation cross section , we first plot in fig .
1 the cross section @xmath34 of eq . ( [ eq3 ] ) as a function of the gluon energy in the @xmath0 rest frame .
it decreases strongly toward the threshold and is broadly peaked around @xmath35 gev , with a maximum value of about 3 mb .
low - momentum gluons have neither the resolution to distinguish the heavy constituent quarks nor the energy to excite them to the continuum . on the other end
, high - momentum gluons also have small cross section with a @xmath0 since they can not see the large size .
we can also express the cross section as a function of the center - of - mass energy of gluons and the @xmath0 , @xmath36 , where @xmath37 .
one can thus translate the energy dependence in fig . 1 into temperature and @xmath7 dependences after thermal average
, since the thermally averaged @xmath38 is proportional to both @xmath7 and temperature @xmath39 . in fig .
2 we plot the thermally averaged gluon-@xmath0 dissociation cross section as a function of temperature for different values of the @xmath0 s transverse momentum @xmath7 .
we observe the same kind of peak structure , with a decreased maximum value due to the thermal average .
the position of the peak also shifts to smaller values of @xmath39 when @xmath7 is increased , corresponding to a fixed value of the averaged center - of - mass energy @xmath40 .
a similar behavior is expected if one plots the thermal cross section as a function of @xmath7 at different temperatures , as done in fig .
3 . however
, in this case , the peak simply disappears at high enough temperatures , because the averaged @xmath40 will be above the threshold value even for @xmath41 .
these features will have considerable consequences for the survival probability of a @xmath0 in an equilibrating parton gas , especially the @xmath7 dependence .
we should also mention that the use of the bose - einstein distribution function has an effect of about 20% on the thermal cross section , compared to that obtained with a boltzmann distribution @xcite .
using the thermal cross section just obtained , we can now calculate the survival probability of @xmath13 in an equilibrating parton plasma . in this paper
, we will neglect the transverse expansion and consider only longitudinal expansion .
we will also only consider @xmath0 suppression in the central rapidity region . a @xmath0 produced at point @xmath42 with velocity @xmath43 in the transverse direction will travel a distance @xmath44 in the time interval @xmath45 before it escapes from a gluon gas of transverse extension @xmath46 ; here @xmath47 .
suppose the system evolves in a deconfined state until the temperature drops below a certain value , which we assume to be 200 mev .
the total amount of time the @xmath0 remains inside a deconfined parton gas is the smaller one of the two times @xmath48 and @xmath49 , the life - time of the parton gas .
assume that the initial production rate of the @xmath0 is proportional to the number of binary nucleon - nucleon interactions at impact - parameter @xmath50 , @xmath51 .
the survival probability of the @xmath0 averaged over its initial position and direction in an equilibrating parton gas is @xmath52 } } { \int d^2r ( r_a^2-r^2 ) } \ ; , \label{eq13}\ ] ] where @xmath53 and @xmath54 is the gluon number density at a given time @xmath55 . in eq . ( [ eq13 ] ) , both the gluon number density @xmath54 and the thermal cross section @xmath56 depend on the temperature , which in turn is a function of time .
in addition , @xmath57 is proportional to gluon fugacity which also evolves with time . to evaluate the survival probability we need to know the entire evolution history of the parton system .
roughly speaking , one can divide this history into two stages : ( 1 ) first there is kinetic thermalization , mainly through elastic scatterings and expansion .
the kinematic separation of free - streaming partons gives us @xmath58 - 0.7 fm/@xmath10 as an estimate of the time when local isotropy in momentum distribution is reached @xcite .
( 2 ) the parton gas now further evolves toward chemical equilibrium through parton proliferation and gluon fusion ; it does so until hadronization or freeze - out , whichever happens first .
this evolution can be determined by a set of master rate equations which give us the time dependence of the temperature and fugacities . in this paper we will not address the question of pre - resonance @xmath0 suppression , which has been discussed and shown to be responsible for the @xmath0 suppression observed in @xmath59 and @xmath60 collisions @xcite .
we will here consider the suppression of fully formed physical @xmath0 states as it should take place if nuclear collision produces dense partonic system .
following ref .
@xcite , we characterize the non - equilibrium of the system by gluon and quark fugacities which are less than unity .
the dominant reactions leading to chemical equilibrium are assumed to be the following two processes : @xmath61 assuming that elastic parton scatterings are sufficiently rapid to maintain local thermal equilibrium , the evolution of the parton densities can be given by the master rate equations . combining these master equations together with one - dimensional hydrodynamic equation
, one can get the following set of equations @xcite : @xmath62 where @xmath63 , @xmath64 , @xmath65 and @xmath66 .
the density and velocity weighted reaction rates @xmath67 can be found in refs .
after taking into account parton screening @xcite and landau - pomeranchuck - migdal effect in induced gluon radiation @xcite , @xmath68 and @xmath69 are found to be functions of only @xmath70 . solving the above rate equations as shown in ref .
@xcite , one finds that the parton gas cools considerably faster than predicted by bjorken s scaling solution @xmath71 = const . ) , because the production of additional partons approaching the chemical equilibrium state consumes an appreciable amount of energy .
the accelerated cooling , in turn , slows down the chemical equilibration process . for the initial conditions given by hijing monte carlo simulations @xcite , at rhic energy the parton system can hardly reach its equilibrium state , since the effective temperature here drops below @xmath72 mev in a time of some 1 - 2 fm/@xmath10 . at lhc energy
, however , the parton gas comes very close to equilibrium , since the system may exist in a deconfined state for as long as 4 - 5 fm/@xmath10 . in the following
we will use the numerical results for the time evolution of the temperature and fugacities obtained from the above master equations to calculate the survival probability of a @xmath0 in such an equilibrating parton system .
shown in fig .
4(a ) are the @xmath0 survival probabilities in the deconfined and equilibrating parton plasma at rhic and lhc energies with initial conditions given by hijing monte carlo simulations @xcite , denoted as set 1 in table [ table1 ] .
we find that there is stronger @xmath0 suppression at lhc than at rhic energy , due both to the higher initial parton densities and longer life - time of the parton plasma .
the increase of the survival probabilities with @xmath0 s transverse momentum is a consequence of the decrease of the thermal cross section with increasing @xmath7 at high temperatures , as shown in fig . 3 , and the shorter time spent by a higher-@xmath7 @xmath0 inside the parton plasma , an effect first considered in ref.@xcite .
@xmath0 dissociation during the late stage of the evolution should have a peak and increase a little with @xmath7 when the temperature drops below 0.3 gev , as illustrated in fig .
3 . this behavior flattens the total survival probability at small values of @xmath7 when we integrate over the entire history of the evolution from high initial temperatures . for a parton system with a low initial temperature ( below 300 mev )
, the @xmath7 dependence of the survival probability should be even flatter .
one can therefore use the @xmath7 dependence to shed light on the initial temperature and the evolution history of the system . to demonstrate the effects of the chemical non - equilibrium in the initial system
, we also show in fig .
4(b ) the survival probabilities in an ideal parton gas with the same initial temperatures as before , but with full chemical equilibrium ( unit initial fugacities ) . in this case
, the temperature simply decreases like @xmath73 .
we see that the @xmath0 is now much more suppressed than in the case of an equilibrating parton plasma , because of both the higher parton density and the longer life time of the system . since there is considerable uncertainty in the estimate of the initial parton production by hijing monte carlo simulations , as discussed in ref .
@xcite , we would like to test the sensitivity of @xmath0 suppression to this .
we therefore multiply initial parton number densities by a factor of 4 , thus increasing initial parton fugacities .
we denote such initial conditions as set 2 in table [ table1 ] .
if the uncertainties in initial conditions are caused by the soft parton production from the color mean fields , the initial effective temperature will decrease .
therefore , we can alternatively increase the initial parton density by a factor of 4 and at the same time decrease @xmath74 to 0.4 and 0.72 gev at rhic and lhc energies , respectively .
this leads to higher initial fugacities , listed as set 3 in table [ table1 ] .
the corresponding survival probabilities calculated with these two sets of initial conditions are shown in fig .
we can see that the @xmath0 suppression is much stronger if the initial parton densities are higher . comparing the solid and dashed lines
, however , shows that the @xmath0 suppression is less sensitive to the variation of the initial temperature and fugacities as far as the parton densities are fixed .
to summarize , we have used the cross section of @xmath13 dissociation by gluons to calculate the @xmath0 suppression in an equilibrating parton gas produced in high - energy nuclear collisions .
the large average momentum in the hot gluon gas enables gluons to break up the @xmath0 , while hadron matter at reasonable temperature does not provide sufficiently hard gluons .
we find a substantial @xmath0 suppression in such a non - equilibrium partonic medium ; however , it is smaller than that in a fully equilibrated parton plasma . in particular , in an equilibrating plasma the behavior of the @xmath0-gluon cross section at high gluon momenta reduces the @xmath0 suppression at large @xmath7 .
in addition to the @xmath0 dissociation during the equilibration of the parton plasma , there are other possible sources of suppression for the actually observed @xmath0 s .
as already noted , nuclear modifications of the @xmath6 production process e.g. , through modified gluon distributions in a nucleus @xcite , multiple scattering accompanied by energy loss @xcite , or a suppression of the nascent @xmath0 before it forms an actual physical resonance @xcite must be taken into account .
such effects would cause @xmath0 suppression in addition to what we have obtained from the equilibrating parton plasma and modify the transverse momentum dependence of @xmath0 suppression @xcite .
moreover , interactions between gluons and @xmath75 bound states before the kinetic thermalization of the partons could also lead to a substantial @xmath0 suppression ; this would depend on the time needed to achieve local momentum isotropy . on the other hand , gluon fusion could also result in @xmath0 production during the evolution of the parton system , similar to the enhancement of open charm .
although studies of pre - equilibrium open charm production indicate @xcite that @xmath0 production during the parton evolution is not significant compared to primary production , a consistent study of @xmath0 suppression should include the pre - equilibrium production in a form of a master rate equation .
x.x thanks the nuclear theory group at lbl berkeley for their hospitality during his visit , when this work was carried out .
d.k , h.s and x .- n.w .
thank k. j. eskola for stimulating discussions .
this work was supported by the u.s .
department of energy under contract no .
de - ac03 - 76sf00098 and by the german research ministry ( bmbw ) under the contract 06 bi 721 .
99 f. karsch , e. laermann , rept .
. phys . * 56 * , 1347 ( 1993 ) ; n. h. christ , nucl . phys . *
a544 * , 81c ( 1992 ) ; y. iwasaki , _ et al .
67 * , 3343 ( 1991 ) ; f. r. brown , _ et al .
lett . * 65 * , 2491 ( 1990 ) .
r. c. hwa and k. kajantie , phys .
lett . * 56 * , 696 ( 1986 ) ; j.p .
blaizot , a.h .
mueller , nucl . phys . *
b289 * , 847 ( 1987 ) .
k. kajantie , p. v. landshoff and j. lindfors , phys . rev
. lett . * 59 * , 2517 ( 1987 ) ; k. j. eskola , k. kajantie and j. lindfors , nucl . phys . *
b323 * , 37 ( 1989 ) ; k. j. eskola , k. kajantie and j. lindfors , phys . lett .
* b214 * , 613 ( 1989 ) .
n . wang and m. gyulassy , phys .
d * 44 * , 3501 ( 1991 ) ; comp .
* 83 * , 307 ( 1994 ) .
k. geiger and b. mller , nucl .
b369 * , 600 ( 1992 ) ; k. geiger , phys .
d * 47 * , 133 ( 1993 ) .
h. j. moehring and j. ranft , z. phys .
c*52 * , 643 ( 1991 ) ; p. aurenche , _ et al .
d * 45 * , 92 ( 1992 ) ; p. aurenche , _ et al .
. phys . commun . *
83 * , 107 ( 1994 ) .
e. shuryak , phys .
lett . * 68 * , 3270 ( 1992 ) ; l. xiong and e. shuryak , phys .
c * 49 * , 2207 ( 1994 ) .
k. geiger and j. i. kapusta , phys .
d * 47 * , 4905 ( 1993 ) .
t. s. bir ' o , e. van doorn , b. m " uller , m. h. thoma and x .-
wang , phys .
c * 48 * , 1275 ( 1993 ) .
j. alam , s. raha and b. sinha , phys .
lett . * 73 * , 1895 ( 1994 ) h. heiselberg and x .-
wang , hep - ph/9504244 .
t. s. bir , b. mller , and x .-
wang , phys .
lett . * b283 * , 171 ( 1992 ) . b. mller and x .-
wang , phys .
lett . * 68 * , 2437 ( 1992 ) ; p. lvai , b. mller and x .-
wang , phys .
c * 51 * , 3326 ( 1995 ) .
k. geiger , phys .
d * 48 * , 4129 ( 1993 ) . z. lin and m. gyulassy , phys .
c * 51 * , 2177 ( 1995 ) .
e. shuryak and l. xiong , phys .
* 70 * , 2241 ( 1993 ) .
m. t. strickland , phys . lett . *
b331 * , 245 ( 1994 ) .
t. matsui and h. satz , phys .
b178 * , 416(1986 ) d. kharzeev and h. satz , phys . lett . *
b334 * , 155(1994 ) m. e. peskin , nucl . phys . * b156 * , 365(1979 ) ; g. bhanot and m. e. peskin , nucl . phys .
* b156 * , 391(1979 ) a. kaidalov , in _ qcd and high energy hadronic interactions , j. trn thanh vn ( ed . ) , edition frontieres , gif - sur - yvette , 1993_. d. kharzeev , l. mclerran and h. satz , phys . lett . *
b356 * , 349(1995 ) .
d. kharzeev and h. satz , cern - th/95 - 117 , bi - tp 95/20 , to appear in _ quark - gluon plasma ii _
, r. c. hwa ( ed . )
( world scientific , singapore ) k. j. eskola and x .-
wang , phys .
d * 49 * , 1284(1994 ) .
j. d. bjorken , phys .
d * 27 * , 140(1983 ) . c. gerschel and j. hfner , z. phys .
c * 56 * , 171 ( 1992 ) .
k. boreskov , a. capella , a. kaidalov and j. trn thanh vn , phys .
d * 47 * , 919 ( 1993 ) .
wong , ornl - ctp-95 - 04 , june , 1995 .
d. kharzeev and h. satz , cern - th/95 - 214 , august , 1995 ; phys . lett .
* b * , to appear .
m. gyulassy and x .-
wang , nucl .
b420 * , 583 ( 1994 ) ; x .-
wang , m. gyulassy and m. pl ' umer , phys .
d * 51 * , 3436 ( 1995 ) .
r. baier et al . , phys . lett . *
b356 * , 349(1995 ) . f. karsch and r. petronzio , phys . lett . *
b193 * , 105(1987 ) .
a. h. mueller and j. qiu , nucl . phys . *
b268 * , 427 ( 1986 ) . k. j. eskola , j. qiu and x
wang , phys .
72 * , 36 ( 1994 ) .
s. gavin and j. milana , phys .
lett . * 68 * , 1834 ( 1992 ) .
s. gavin , m. gyulassy and a. jackson , phys .
b208 * , 257 ( 1988 ) ; s. gavin and m. gyulassy , phys . lett . * b214*,241 ( 1988 ) . | short - distance qcd is employed to calculate the @xmath0 survival probability in an equilibrating parton gas , whose evolution is governed by a set of master rate equations .
partons in the early stage of high - energy nuclear collisions may initially not be in equilibrium , but their average transverse momentum is sufficiently high to break up a @xmath1 bound state .
such a break - up during the evolution of the parton gas is shown to cause a substantial @xmath0 suppression at both rhic and lhc energies , using realistic estimates of the initial parton densities .
the transverse momentum dependence of the suppression is also shown to be sensitive to the initial conditions and the evolution history of the parton plasma . 6.0 in 9.0 in |
molecular electronics is an active field with many potential applications and novel innovations in nanoelectronic devices @xcite .
an ultimate goal in molecular electronics is to construct molecule - based integrated circuits . to achieve this goal , understanding electron transport through a single - molecule device ( junction ) is a required first step . during the past two decades
, great experimental progress has been made on single - molecule junctions due to advancements in microfabrication and self - assembly techniques @xcite .
in addition , improvements in theoretical modeling have enabled successful explanations and predictions of experimental observations , e.g. , kondo effect@xcite , thermoelectricity @xcite , orbital gating @xcite , destructive quantum interference @xcite , and conductance dependence upon molecular conformation @xcite .
the landauer formula combined with the non - equilibrium green s function method @xcite has become a common approach to analyze electron transport through a single - molecule junction in the absence of applied fields . for a practical device
, the capability of manipulating current by applying such fields is necessary .
therefore , control of electron transport , e.g. , by means of light @xcite or gate electrodes @xcite , is a prime issue for molecular electronics .
light is a potential control tool for operating ultrafast electronic devices due to a wide variety of control options , e.g. , field strength , phase , frequency , and polarization , compared with a static field provided by gate electrodes . since 1960s @xcite electron transport in the presence of a coherent light field
has attracted general interest due to photon assisted tunneling observed in mesoscopic systems exposed to microwave radiation @xcite . in superconductor - insulator - superconductor tunnel junctions
@xcite and semiconductor nanostructures @xcite , light - driven electron transport has been extensively studied for some time .
recently , light - driven electron transport at the molecular level became an emerging field due to experimental improvements in molecular junctions @xcite .
moreover , a wealth of coherent light sources in the visible and the infrared regime , compatible with electronic energy scales of molecular systems , motivated innovative theoretical studies in this field @xcite . within a time - dependent perturbation theory along with the extended hckel model , photon - assisted tunneling was predicted in the presence of high - intensity fields @xcite .
theoretical studies using floquet theory together with a tight - binding model have shown that molecular wires can act as coherent quantum ratchets @xcite , and that current rectification @xcite , shot noise control @xcite , photon - assisted tunneling @xcite , and coherent destruction of tunneling could be observed in a linear molecular wire .
vibrational effects @xcite , memory effects @xcite , and optimal control of shot noise @xcite have been investigated in a two - level model .
however , most studies focus on linear molecular wires or few - level models and neglect the nature of molecular structure , e.g. , cross - conjugated molecules and polycyclic aromatic hydrocarbons , which offer multiple pathways for tunneling electrons . multiple pathways in a molecular junction can be regarded as forming a `` network '' , which is of importance in molecular electronics for several reasons .
first , such networks can be thought of as analogous to electric circuits , e.g. , a double - backbone molecule corresponding to a parallel circuit . based on kirchhoff s circuit laws ,
the conductance of two parallel constituents in a circuit equals the sum of the conductance of the individuals .
however , at the molecular scale kirchhoff s circuit laws do not hold since the phase coherence length of tunneling electrons is comparable to the circuit scale , i.e. , quantum interference effects play an important role .
a conductance superposition law in a parallel circuit has been proposed @xcite and experimentally reported @xcite .
second , pathway differences in a network can lead to destructive quantum interference @xcite , which can be exploited to form single - molecule optoelectronic switches @xcite , single - molecule electric revolving doors @xcite , and quantum interference effect transistors @xcite .
destructive quantum interference with tunneling has been experimentally observed in aromatic hydrocarbons @xcite and cross - conjugated molecules @xcite at room temperature , indicating that molecular structure can be a key resource for coherent quantum transport .
third , a network such as a single molecule connected to multi - terminals @xcite could form a nanoscale router . in order to manipulate current in a molecular router , an understanding of the correlation between the molecular structure and current direction
is required .
we present a new general formulation to explore laser - driven transport through three individual networks consisting of ( i ) a two - terminal molecular junction based on phenyl - acetylene macrocycle ( pam ) @xcite , ( ii ) pam with three terminals , and ( iii ) a parallel molecular circuit .
the formulation generalizes a scattering analysis based on single - particle green s functions within the framework of the non - hermitian floquet theory @xcite .
the new formulation is valid for arbitrary strength as well as frequency of a driving field and enables the modeling of light - driven transport through multi - terminal and multi - orbital systems in terms of transmission probabilities .
we will discuss field - amplitude power laws in the weak - field regime , the effects of asymmetric molecule - electrode couplings , and the influence of laser polarization on electron transport through pam .
electron transport through a single - molecule junction in the presence of time - periodic fields can be described by the time - dependent hamiltonian @xmath0 composed of the external - field - driven molecular hamiltonian @xmath1 , the lead ( electrode ) hamiltonian @xmath2 , and the molecule - lead coupling term @xmath3 , i.e. , @xmath4 . to focus on the effect of molecular networks , we remove other possible effects , e.g. , many - body interactions , and adopt a single - electron tight - binding model to describe @xmath1 @xmath5 where @xmath6 ( @xmath7 ) are fermion operators which annihilate ( create ) an electron in the atomic orbital @xmath8 in the molecule .
the field - driven molecular hamiltonian satisfies @xmath9 due to the time - periodic field with frequency @xmath10 .
note that the field is not limited to lasers , and it can be any type of time - periodic resource . in eq .
( [ model_hamiltonian1 ] ) the atomic orbitals @xmath11 and @xmath12 need not be nearest neighbors , i.e. , long - range interactions between the atomic orbitals @xmath11 and @xmath12 can be considered .
we use a noninteracting electron gas model to describe the leads @xmath13 where @xmath14 is the total number of the electrodes and @xmath15 ( @xmath16 ) are fermion operators which annihilate ( create ) an electron in the state @xmath17 with energy @xmath18 in the lead @xmath19 .
the molecule - lead coupling term is @xmath20 where @xmath21 denotes the electron transfer integrals between the atomic orbital @xmath8 of the molecule and the orbital @xmath17 in the lead .
the coupling function in the energy domain is defined as @xmath22 and its fourier transform in the time domain is @xmath23 the off - diagonal terms of the coupling function originate from the interactions between @xmath17 in the lead @xmath19 and two different atomic orbitals on the molecule . as @xmath17 only interacts with a specific atomic orbital @xmath24 on the molecule , i.e. , @xmath25 , we can derive @xmath26 , consistent with the previous result @xcite . in the heisenberg picture ,
the time - evolution operator @xmath27 associated with the total system hamiltonian @xmath0 is governed by the equation @xmath28 and the equations of motion of the electron annihilation operators can be expressed in terms of @xmath0 and @xmath27 as follows @xmath29 u(t , t_0 ) , \\
\label{heisenberg3_2 } i\hbar\frac{d b_{\alpha q}(t)}{dt}=-u^{\dagger}(t , t_0)[h_{\textrm{tot}}(t),b_{\alpha q}]u(t , t_0).\end{aligned}\ ] ] by substitution of eqs .
( [ model_hamiltonian1 ] ) ( [ model_hamiltonian3 ] ) into eqs .
( [ heisenberg3_1 ] ) and ( [ heisenberg3_2 ] ) , we obtain equations of motion for electron annihilation operators in a single - molecule junction @xmath30 the linear differential equations in eq .
( [ heisenberg4 ] ) have the formal solution @xcite @xmath31 where @xmath32 is determined by the initial conditions @xmath33 and @xmath34 .
the electrodes are initially in a thermal equilibrium state described by the initial density matrix @xmath35 , where @xmath36 is the temperature , @xmath37 is an electron number operator , and @xmath38 is the chemical potential associated with the lead @xmath19 . as a result , the average electron occupation number at @xmath39 can be expressed as @xmath40 , where @xmath41 is the fermi function of the lead @xmath19 .
for a two terminal system with symmetric chemical potentials , the left and right chemical potentials can be respectively written as @xmath42 and @xmath43 , in terms of the zero - bias chemical potential @xmath44 , electric charge e , and the source - drain voltage @xmath45 . to explore electric current in the presence of a field with time - period @xmath46
, we define the time - averaged current @xmath47 in the lead @xmath19 as an integral over the time interval @xmath46 , @xcite @xmath48 the symbol @xmath49 denotes ensemble average , e.g. , @xmath50 is the ensemble - averaged current in the lead @xmath19 at time @xmath51 @xcite , @xmath52 where the ensemble - averaged electron number change @xmath53 between time @xmath51 and @xmath39 is @xmath54 eq . ( [ elec_current ] ) indicates that as the ensemble - averaged electron number in the lead @xmath19 decreases with time , i.e. , @xmath55 , the current flows into the electrode @xmath19 , i.e. , @xmath56 . at the initial time
@xmath57 all electrons are at equilibrium , i.e. , @xmath58 , @xmath40 .
in addition , we assume that at @xmath57 the electrons between the lead and the molecule have no interaction , i.e. , @xmath59=0 . by virtue of these conditions , substituting eq .
( [ fundamental ] ) into eq .
( [ charge1 ] ) results in the relation @xmath60 here the backscattering term has been eliminated by invoking the relation @xmath61 and @xmath62 . after substituting eq .
( [ charge3 ] ) into eq .
( [ elec_current ] ) and taking the long - time limit @xmath63 , the ensemble - averaged current @xmath50 can be expressed as @xmath64 \nonumber \\ & - \textrm{e } \sum_{\alpha'\neq\alpha}\sum_{qq'}\left [ k_{\alpha q,\alpha ' q'}(t)f_{\alpha'}(\epsilon_{\alpha ' q'})-k_{\alpha ' q',\alpha q}(t)f_{\alpha}(\epsilon_{\alpha q})\right ] , \end{aligned}\ ] ] where @xmath65 denotes the rate that an electron tunnels from the orbital @xmath17 in the lead @xmath19 to the orbital @xmath8 on the molecule , and @xmath66 denotes the rate that an electron tunnels from the orbital @xmath67 in the lead @xmath68 to the orbital @xmath17 in the lead @xmath19 . in the long - time limit @xmath63 ,
all transient currents die out so @xmath69 can be ignored in our analysis .
in addition , @xmath70 corresponds to a periodic charging of the molecule driven by external time - periodic fields @xcite and it contributes zero current over a period @xmath46 ( the detailed derivation is in appendix [ appendix1 ] ) .
therefore , the time - averaged current in lead @xmath19 reads @xmath71,\end{aligned}\ ] ] indicating that the current is only relevant to the lead - to - lead tunneling rates and the fermi functions .
furthermore , using the following relations , @xmath72 , \\
\frac{d\langle\alpha q|u(t , t_0)|\alpha ' q'\rangle}{dt}&=-\frac{i}{\hbar}[\epsilon_{\alpha q}\langle\alpha q|u(t , t_0)|\alpha ' q'\rangle \nonumber \\ & + \sum_n v_{\alpha q , n}\langle n|u(t , t_0)|\alpha'q'\rangle],\end{aligned}\ ] ] the lead - to - lead tunneling rate @xmath73 can be expressed in terms of @xmath74 and @xmath75 as @xmath76 to derive the ensemble - averaged current @xmath50 , the evaluation of the matrix elements of the propagators , e.g. , @xmath74 and @xmath75 , is needed .
we start from the interaction picture and separate the total hamiltonian into the uncoupled hamiltonian @xmath77 and the coupling hamiltonian @xmath78 .
the propagator @xmath79 associated with @xmath80 is governed by the equation @xmath81 it is readily seen that @xmath82 , where @xmath83 and @xmath84 . here , @xmath85 is the time ordering operator . by virtue of eqs .
( [ evolution2_1 ] ) and ( [ evolution2 ] ) , we can obtain @xmath86 by substituting eq .
( [ evolution3 ] ) into @xmath87 , @xmath88 and @xmath89 , and using the fact that @xmath90 and @xmath91 , we find @xmath92 eq .
( [ evolution6 ] ) and eq .
( [ evolution5x ] ) can be further simplified using the retarded and advanced green s functions defined as @xmath93 where @xmath94 is the heaviside function and eq .
( [ green1 ] ) satisfies @xmath95^{\dagger}$ ] and @xmath96 .
note that @xmath97 and @xmath98 are the green s functions for the total system . from eq .
( [ green1 ] ) , the fourier transforms of the retarded and advanced green s functions in the energy domain are , respectively , @xmath99 @xmath100 in which @xmath101^{\dagger}$ ] . note that the total hamiltonian has time - periodic symmetry @xmath102 , leading to @xmath103 and @xmath104 . as a result
, the retarded and advanced green s function in the energy domain can be expanded in a fourier series @xmath105 where the fourier coefficient @xmath106 is @xmath107 it can be readily shown that by letting @xmath63 , @xmath108 , and @xmath109 in eq .
( [ evolution6 ] ) , and making use of eq .
( [ green2_1 ] ) , eq .
( [ evolution6 ] ) becomes @xmath110 where @xmath111 stands for @xmath112 for convenience .
similarly , by letting @xmath63 , @xmath113 and @xmath114 , eq .
( [ evolution5x ] ) becomes @xmath115 furthermore , by letting @xmath108 and @xmath109 in eq .
( [ rate_green3 ] ) , the matrix element @xmath89 becomes @xmath116 note that we do not show the term @xmath117 in eq .
( [ rate_green3 ] ) and eq .
( [ rate_green4 ] ) because of the condition @xmath118 in eq .
( [ elec_current01 ] ) . eq .
( [ rate_green2 ] ) and eq .
( [ rate_green4 ] ) can be used for expressing @xmath119 and @xmath73 in eq .
( [ elec_current01 ] ) in terms of the retarded ( advanced ) green s functions . in most of the literature related to coherent quantum transport , the current formula
is expressed in terms of transmission functions . in this section
, we will show that the rate equation ( eq . ( [ elec_current01 ] ) ) and
the lead - to - lead tunneling rates correspond to the landauer - type formula and transmission functions , respectively .
use of @xmath120 and substitution of eq .
( [ rate_green2 ] ) and eq .
( [ rate_green4 ] ) into eq .
( [ rate1 ] ) give @xmath121^*g^{\textrm{r}}_{n_3n_4}(t-\tau,\epsilon_{\alpha ' q ' } ) + c.c.,\end{aligned}\ ] ] where we use @xmath122 and @xmath123 instead of @xmath11 and @xmath12 in eq .
( [ rate_green2 ] ) as well as @xmath124 and @xmath125 instead of @xmath11 and @xmath12 in eq .
( [ rate_green4 ] ) . by invoking eq .
( [ rate3 ] ) and after some manipulations , @xmath126 can be written as @xmath127^*g^{\textrm{r}}_{n_3n_4}(t-\tau,\epsilon_{\alpha ' q ' } ) f(\epsilon_{\alpha'q ' } ) + c.c.\end{aligned}\ ] ] by using eq .
( [ self1 ] ) , eq .
( [ self2 ] ) and @xmath128 , eq .
( [ rate4 ] ) becomes @xmath129^*g^{\textrm{r}}_{n_3n_4}(t-\tau,\epsilon ) f(\epsilon ) + c.c.\end{aligned}\ ] ] then , by use of eq .
( [ green_fourier ] ) , expansion of @xmath130^*$ ] and @xmath131 in eq .
( [ rate5 ] ) in a fourier series gives @xmath132^*g_{n_3n_4}^{\textrm{r}(k)}(\epsilon ) f_{\alpha'}(\epsilon)+c.c.\end{aligned}\ ] ] since the complex conjugate part in eq .
( [ rate6 ] ) contributes @xmath133 , we can make use of the fourier transform of the coupling function , i.e. , @xmath134 , and reduce eq .
( [ rate6 ] ) to @xmath135 f_{\alpha'}(\epsilon ) , \end{aligned}\ ] ] where @xmath136=\sum_{n_1n_2n_3n_4 } \gamma_{\alpha , n_1 n_3}(\epsilon+k\hbar\omega)g_{n_3n_4}^{\textrm{r}(k)}(\epsilon)\gamma_{\alpha',n_4 n_2}(\epsilon)g_{n_2n_1}^{\textrm{a}(k')}(\epsilon)$ ] .
@xmath137 can be derived by repeating the procedure from eq .
( [ rate4 ] ) to eq .
( [ rate7 ] ) . finally , substitution of eq .
( [ rate7 ] ) into eq .
( [ elec_current01 ] ) and use of @xmath138 give the time - averaged current without spin degeneracy @xmath139,\end{aligned}\ ] ] and the transmission functions @xmath140,\end{aligned}\ ] ] which corresponds to the tunneling of an electron from the lead @xmath68 to the lead @xmath19 with energy @xmath141 accompanied by k - photon absorption ( @xmath142 ) or emission ( @xmath143 ) .
note that in general cases @xmath144 if molecules have no generalized parity symmetry @xcite .
( [ elec_current02 ] ) enables dealing with transport in multi - terminal systems in a time - periodic driving field and it is straightforward to show that @xmath145 which satisfies the continuity equation . the correspondence between eq .
( [ elec_current01 ] ) and eq . ( [ elec_current02 ] ) reveals the connection between the lead - to - lead tunneling rates and the transmission functions .
in addition , the trace form in eq .
( [ transmission ] ) enables the formulation to be applied to realistic models , e.g. , a molecular hamiltonians derived from the density - functional method in a maximally localized wannier function representation @xcite . in the absence of the external driving field ( @xmath1 is time - independent ) , using the fact that @xmath146 reduces eq .
( [ elec_current02 ] ) to the landauer - type current formula , @xmath147,\end{aligned}\ ] ] where @xmath148 $ ] , which is consistent with the results derived from fisher - lee relation @xcite and meir - wingreen formula @xcite .
for a two terminal system , i.e. , @xmath149 r and @xmath150 l , assuming that the coupling function satisfies @xmath151 and @xmath152 , eq .
( [ elec_current02 ] ) reduces to @xmath153,\end{aligned}\ ] ] where @xmath154 and @xmath155 , which are consistent with the results derived by hnggi et al .
@xcite . to solve for the fourier coefficients of the retarded ( advanced ) green s functions , we start from the schrdinger picture and consider the dynamics of @xmath157 , @xmath158_{nn '' } \langle n''|u(t , t_0)|n ' \rangle \nonumber \\ & + \sum_{\alpha q } v_{\alpha q , n}^ * \langle \alpha q|u(t , t_0)|n ' \rangle.\end{aligned}\ ] ] the first and second terms on the right - hand side correspond to the dynamics governed by the molecular hamiltonian and influenced by the leads , respectively . by virtue of eq .
( [ self2 ] ) and eq .
( [ evolution3 ] ) , the last term in eq .
( [ sch_pro2 ] ) can be expressed in terms of memory functions as @xmath159 moreover , we adopt the markov process approximation for the memory function , i.e. , @xmath160 , and the last term in eq .
( [ sch_pro2 ] ) becomes @xmath161 where the factor of @xmath162 comes from @xmath163 .
note that the markov process approximation for the memory function is equivalent to the wide band limit approximation , i.e. , @xmath164 . as @xmath165 is energy - independent ,
its fourier transform is @xmath166 .
substitution of eq .
( [ sch_pro4 ] ) into eq .
( [ sch_pro2 ] ) gives @xmath167_{nn '' } -\frac{i}{2 } \sum_{\alpha}\sum_{n '' } \overline{{\gamma}}_{\alpha , nn '' } \right)\langle n''|u(t , t_0)|n ' \rangle,\end{aligned}\ ] ] which can be equivalently written as @xmath168 where the wavefunction @xmath169 is related to the propagator @xmath27 via the relation @xmath170 .
( [ sch_pro6 ] ) is a linear , time - periodic ( @xmath9 ) , non - hermitian ordinary differential equation , whose complete solution based on the floquet theory can be written as @xmath171 $ ] and @xmath172 are the floquet exponent ( characteristic exponent ) and floquet eigenstates respectively .
moreover , @xmath173 and @xmath174 correspond to the quasienergy and lifetime of the floquet eigenstates @xmath172 . here
the floquet eigenstate is a periodic function of time , i.e. , @xmath175 . by substitution of eq .
( [ floquet1 ] ) into eq .
( [ sch_pro6 ] ) , the underlying time - dependent schrdinger equation can be cast into the floquet eigenvalue equation @xmath176 where @xmath177 is the so - called floquet hamiltonian in an extended hilbert space @xcite .
note that @xmath177 is non - hermitian hamiltonian @xcite so its adjoint eigenvalue equation satisfies @xmath178 the eigenstates @xmath172 and the adjoint states @xmath179 form a complete biorthogonal basis in an extended hilbert space .
due to the time - periodic symmetry of the floquet eigenstates @xmath175 , a complete solution can be also constructed as @xcite , @xmath180 for any integer @xmath181 and @xmath182 . for consistency , we use the notation @xmath183 instead of @xmath172 in the following derivations .
the quasienergies @xmath184 can be mapped into the first brillouin zone , @xmath185 , where @xmath186 is an arbitrary chosen real number .
note that for fixed time @xmath51 , the floquet states of the first brillouin zone form a complete set in _ r _ , i.e. , @xmath187 , where 1st bz denotes the first brillouin zone @xcite . by substituting eq .
( [ floquet3 ] ) into eq .
( [ sch_pro6 ] ) , we arrive at the following eigenvalue equation @xmath188 where @xmath189 has time - periodic symmetry and can be decomposed into a fourier series , @xmath190 and eq . ( [ floquetcoef_2 ] ) into eq .
( [ floquet4 ] ) , and perform the fourier transform on both sides of eq .
( [ floquet4 ] ) , i.e. , @xmath191 , and then derive a time - independent infinite - dimensional eigenvalue matrix equation @xmath192_{n'k',nk}\phi^{n , k}_{\lambda,\zeta } = q_{\lambda\zeta}\phi^{n',k'}_{\lambda,\zeta}.\end{aligned}\ ] ] here , @xmath193 and @xmath194_{n'k',nk}$ ] denotes the matrix elements of the time - averaged floquet hamiltonian over a period @xmath46 .
moreover , @xmath194_{n'k',nk}$ ] can be cast as @xmath195_{n'k',nk } & = [ h^{(k'-k)}_{\textrm{mol}}]_{n'n}-\delta_{k'k}\frac{i}{2}\sum_{\alpha } \overline{{\gamma}}_{\alpha , n'n } \nonumber \\ & + \delta_{n'n}\delta_{k'k } k\hbar\omega,\end{aligned}\ ] ] where @xmath196 is the fourier coefficient of @xmath1 defined as @xmath197 similarly , we can derive the adjoint eigenvalue equation of eq .
( [ fm_adj ] ) @xmath198_{n'k',nk}\chi^{n , k}_{\lambda,\zeta } = \tilde{q}_{\lambda\zeta}\chi^{n',k'}_{\lambda,\zeta}.\end{aligned}\ ] ] with the help of eq .
( [ floquet1 ] ) , it is straightforward to show that the propagator in eq .
( [ sch_pro2 ] ) can be expressed as @xmath199 by combining eq .
( [ green2_1 ] ) , eq .
( [ green_fourier2 ] ) and eq .
( [ evolution_nonhermitian ] ) , and making use of eq .
( [ floquetcoef_3 ] ) , the fourier coefficients of the retarded green s function can be written as @xmath200 where @xmath201 @xcite .
note that eq .
( [ green2 ] ) is used to numerically solve for the retarded green s functions in this paper . as @xmath202 is time - independent , @xmath203 for @xmath204 . as a result ,
( [ green2 ] ) can reduce to @xmath205 which is equivalent to the retarded green s function @xmath206 derived from a time - independent hamiltonian eigenvalue matrix equation @xmath207 for a two - level system it is more convenient to use eq .
( [ green_zerofield2 ] ) instead of eq .
( [ green_zerofield ] ) since the former can be solved analytically ( see section [ section4 - 2 ] ) .
to summarize , current through a single - molecule junction under a time - periodic field can be computed numerically according to the following five steps : ( i ) construct the molecular hamiltonian @xmath1 and coupling functions @xmath208 , ( ii ) solve the floquet eigenvalue equation , eq .
( [ fm ] ) , and then obtain its eigenvalues and eigenvectors , ( iii ) evaluate the matrix element of the retarded green s function by substitution of the eigenvalues and eigenvectors into eq .
( [ green2 ] ) , ( iv ) compute the transmission functions by using eq .
( [ transmission ] ) , and ( v ) compute the current by using eq .
( [ elec_current02 ] ) .
is the angle between the polarization direction of the laser and the x axis.,width=302 ] here and in section [ section4 ] , we explore laser - driven transport through three systems with multiple pathways by employing the method derived in the section [ section2 ] . the first system shown in fig .
[ fig1 ] ( a ) is a two - terminal molecular junction based on a pam molecule @xcite in the presence of a monochromatic laser field . according to a previous study @xcite
, the first system is an effective optoelectronic switching device due to its large on - off current ratios and weak - field operating conditions , both of which are not sensitive to the fermi level of the electrodes .
however , except for the fermi level of the electrodes , the previous study did not consider other experimental conditions , e.g. , asymmetric molecule - lead coupling and laser polarization , which may affect the large on - off current ratios and weak - field operation . to examine whether the first system is a robust optoelectronic switch
, we study laser - driven transport through the pam molecule by changing the field amplitude , coupling function strength , and laser polarization direction . for simplicity
, we adopt the hckel model and the electric dipole approximation to describe the molecular hamiltonian @xmath1 because pam is a typical conjugated molecule and we only consider weak - field operation . as a result
, the molecular hamiltonian can be expressed as @xmath209 where @xmath210 is the energy of the @xmath211-orbital on the carbon atoms , @xmath212 is the resonance integral between directly - bonded carbon atoms , @xmath213 and @xmath214 are fermion operators which create and annihilate an electron in the @xmath215-orbital @xmath8 on the n - th carbon of pam at position @xmath216 , and @xmath217 is a time - dependent electric field propagating along the z direction with the field amplitude e , frequency @xmath218 , and polarization angle @xmath219 between the laser field and the x axis ( see fig .
[ fig1 ] ( b ) ) .
note that @xmath220 is derived from the geometry optimization of pam at the b3lyp/6 - 31 g(d ) level using the gaussian 09 program @xcite .
the parameters @xmath221 ev and @xmath222 ev are from photo - electron spectroscopy experiments @xcite . within the wide band limit approximation ,
the coupling functions have the form @xmath223 where @xmath224 and @xmath24 respectively denote the @xmath211-orbital on the contact carbon atoms @xmath225 and @xmath226 . substituting eq .
( [ molhamiltonian ] ) into eq .
( [ fcoefficeint ] ) gives @xmath227 substitution of eqs .
( [ coupmatrix1_1 ] ) , ( [ coupmatrix1_2 ] ) , and ( [ molhamiltonian2 ] ) into eq .
( [ fm1 ] ) yields the following time - averaged floquet hamiltonian matrix @xmath228 where @xmath229 and @xmath230 are block matrices with elements @xmath231_{n',n}-(i/2)(\delta_{n'u}\delta_{nu}\overline{\gamma}_{\textrm{l},uu}+\delta_{n'v}\delta_{nv}\overline{\gamma}_{\textrm{r},vv})$ ] and @xmath232_{n',n}$ ] , respectively . by solving eq .
( [ fm ] ) and performing steps ( iii ) ( v ) , we can obtain the transmission and the light - driven current through the pam - based optoelectronic switch . in an experiment ,
the measured current through a single - molecule junction is dependent on the chemical nature of the electrodes and linker groups , e.g. , the types of electrodes such as au and ag @xcite , electrode conformations such as au(111 ) and au(100 ) , hollow and on - top contacts @xcite , and the types of linker groups such as thiol ( @xmath233sh ) and amine ( @xmath234 ) groups @xcite .
the influence of these effects can be modeled as the self - energy derived from scattering formulations or non - equilibrium green s functions , if the system is in the coherent tunneling regime , i.e. , molecules with short length and large injection gap at low temperature @xcite . therefore , in this section , we model the influence of the electrodes and the linker groups via changing the values of the coupling functions ( the imaginary term of the self - energy ) in order to examine whether the destructive quantum interference caused by the pam - based molecular network is sensitive to molecule - lead couplings .
[ fig2 ] depicts the transmission of the two - terminal pam junction with different molecule - lead coupling strength in the absence of a laser field . for @xmath235 , @xmath236 , @xmath237 ev ( we consider symmetric molecule - lead couplings , i.e. , @xmath238 )
, the transmission is strongly suppressed at energies @xmath239 @xcite .
thus , the transmission characteristics are not sensitive to the values of the coupling functions .
the transmission suppression originates from destructive quantum interference resulting from the meta - connected benzene unit , indicated by the dashed lines in fig .
[ fig1 ] ( b ) , and the repeated meta - connected benzene units can suppress the transmission of a tunneling electron and broaden the range of anti - resonance @xcite .
the values of the coupling functions do not strongly influence the transmission characteristics because the meta - connected benzene units are not directly coupled to the two leads .
in addition , the wide anti - resonance range indicates that the current is minute in the small source - drain voltage limit , e.g. , @xmath240 ampere at @xmath241 and @xmath242v . note that we do not include the spin degeneracy for the computed current . as a result ,
a two - terminal pam junction in the absence of a laser field can function as the `` off - state '' of a single - molecule switch because its extremely small transmission is robust to the change of molecule - lead couplings and the fermi level of the electrode .
all transmission peaks in fig .
[ fig2 ] occur in pairs with the center @xmath243 because pam is an alternant hydrocarbon @xcite .
in addition , the molecular orbital energies of pam are @xmath244 ev , @xmath245 ev , @xmath246 ev , @xmath247 ev , @xmath248 ev , and @xmath249 ev , which are in good agreement with the six transmission peaks in fig . [ fig2 ] .
we do not consider the real part of the self - energy in our analysis because it just shifts the energy of resonant states and does not influence the transmission characteristics .
, where the abscissa is the energy of the tunneling electron.,width=302 ] v , @xmath250 ev , @xmath251 , @xmath252 , @xmath253 , and @xmath254.,width=302 ] consider the following experimental conditions for a two - terminal pam junction .
we assume a small source - drain voltage ( @xmath242v ) , low temperature limit ( @xmath250 ev ) , and symmetric chemical potentials ( @xmath253 and @xmath254 ) .
[ fig3 ] shows that the current - field intensity characteristics of a two - terminal pam junction .
the solid lines are proportional to the second power of the field amplitude @xmath255 in the range @xmath256 v / cm and the dashed lines are proportional to the fourth power of @xmath255 in the range from @xmath257 to @xmath258 v / cm .
this phenomenon is due to photon - assisted tunneling , which occurs under the condition that an integer multiple of laser frequencies are compatible with the energy difference between the resonant - state energies of a two - terminal pam junction and the fermi level of electrodes , i.e. , @xmath259 . consider the case of one - photon assisted tunneling ( @xmath260 ) , the frequency @xmath261 is @xmath262 ev @xmath263ev , consistent with the red solid line in fig .
[ fig3 ] . as a result
, we can conclude that the solid lines and the dashed lines correspond to current induced by one - photon assisted tunneling and two - photon assisted tunneling , respectively
. the field - amplitude power laws of current induced by photon assisted tunneling can be understood by using time - independent , non - hermitian perturbation theory @xcite in appendix [ appendix2 ] .
as the electric dipole term in @xmath1 is sufficiently weak , appendix [ appendix2 ] shows that the one - photon green s function @xmath264 and the two - photon green s function @xmath265 are proportional to @xmath255 and @xmath266 , respectively . the power laws
do not hold for @xmath267 v / cm because of stark shifting of the quasi - states .
moreover , we can observe the deviation of the dashed lines from the power laws for a very weak field , i.e. , @xmath268 v / cm , which results from the overlap of one - photon and two - photon assisted tunneling .
therefore , by use of photon - assisted tunneling , a two - terminal pam junction can function as the `` on - state '' of a single - molecule switch .
moreover , a pam - based switch may be used for a wide range of applications due to its weak - field operating conditions .
[ fig3 ] shows the effective range of the field amplitude power laws for the pam - based optoelectronic switch , indicating that the power laws may be applied to examine the one - photon and two - photon assisted tunneling in a molecular junction experiment .
although great progress has been made in fabrication techniques over the past thirty years , it is still not easy to experimentally create a high - quality symmetric molecular junction .
the effects of asymmetric molecular junctions on electron transport in the presence of a driving field are non - trivial since they can lead to the coherent quantum ratchet effect @xcite and current rectification @xcite , which may alter the large on - off current ratios of the pam - based switch . in order to clarify these effects ,
we explore laser - driven transport through a two - terminal asymmetric molecular junction with different frequencies .
v , @xmath250 ev , @xmath252 , @xmath269 ev , @xmath270 ev , @xmath253 , and @xmath254.,width=302 ] based on the argument in section [ section3 - 1 ] , the asymmetric effects arising from the electrodes and linker groups can be modeled as molecule - lead coupling functions with different values , @xmath271 .
[ fig4 ] shows that the peaks of all lines are located at the same frequencies ( @xmath272 , @xmath273 , and @xmath274 ev ) , indicating that the existence of these peaks caused by one - photon assisted tunneling do not change with asymmetric molecule - lead couplings .
the asymmetric molecule - lead couplings only affect the peak heights and widths .
note that our analysis is based on the wide band limit approximation , i.e. , we do not consider the effect of the real part of self - energy .
however , for a large molecule , such as pam , the energy shifting caused by the intermediate couplings should not significantly change the electronic properties . as a result , we conclude that the large on - off current ratios of the pam - based switch should be robust to asymmetry in the molecular junction .
the field polarization can play a crucial role in the control of electron dynamics .
recently , several theoretical and experimental studies have shown that the aromaticity of benzene @xcite , aromatic ring currents in ring - shaped molecules @xcite , and electron localization in molecular dissociation @xcite can be manipulated by changing the laser polarization . however , these studies focus on the influence of strong linearly or circularly polarized light on isolated molecules , not a molecular junction , which motivates us to explore electron transport through a molecular junction in a weak laser field with linear polarization in various directions . in a two - terminal pam junction for @xmath242
v , @xmath250 ev , @xmath269 v / cm , @xmath275 ev , @xmath253 , and @xmath254.,width=302 ] fig . [ fig5 ] depicts the current through a two - terminal pam junction in a weak field with different frequencies as a function of laser polarization angle @xmath219 .
the light with the frequencies @xmath276 , @xmath273 , and @xmath274 ev can result in one - photon assisted tunneling .
all of the results show the same characteristics : the current reaches the minimum at the laser polarization angle @xmath277 and the maximum at @xmath252 and @xmath278 .
the current maximum occurs at @xmath252 and @xmath278 because the oscillation of the electric field can assist electron tunneling when the directions of electron transport and laser polarization are the same , i.e. , along the x - direction in fig .
[ fig1 ] . on the other hand ,
when the directions of electron transport and laser polarization are vertical , i.e. , @xmath277 , the oscillation of the electric field can not assist electron tunneling , leading to the minimum electric current .
due to all of the curves in fig .
[ fig5 ] with similar structure , we define the time - averaged current as @xmath279 in a laser field with @xmath218 and at @xmath252 , and the factor @xmath280 describes the relation between current and laser polarization angle , @xmath281 fig .
[ fig6 ] shows that all normalized current curves @xmath282 coincide @xmath283 , i.e. , @xmath284 the independence of @xmath280 upon frequency is interesting and deserves further exploration . the @xmath283 behavior can be understood by the perturbation analysis in appendix [ appendix2 ] . from eq .
( [ h1_element ] ) , the one - photon green s function @xmath264 is proportional to @xmath285 , because in a weak field only the @xmath286 component of the electric field can assist electron tunneling and the @xmath287 component can not .
therefore , the cosine squared relation comes from @xmath288 . according to the results shown in fig .
[ fig5 ] , we can conclude that the on - off switching ratios are sensitive to the laser polarization , indicating that the current magnitude can be conveniently manipulated this way in the laboratory . in a two - terminal pam junction.,width=302 ]
our methodology is also applicable to multi - terminal systems . to illustrate the methodology , we give two examples , a single - molecule optoelectronic router and a molecular parallel circuit .
the former is a three - terminal system connected to three electrodes and the latter is a four - terminal system connected to two electrodes . the molecular networks exhibit a variety of novel and interesting physical phenomena . inspired by an analogy from electrical engineering , we propose a new type of current router and explore the correlation between the direction of laser polarization and current . a single - molecular optoelectronic router as shown in fig .
[ fig7 ] composed of a pam molecule connected to three electrodes in the presence of a monochromatic laser field .
the hamiltonian is described in eqs .
( [ molhamiltonian ] ) and ( [ polarization ] ) , and we assume the coupling functions are @xmath289 , @xmath290 , and @xmath291 , where @xmath224 , @xmath24 , and @xmath292 denote the @xmath211-orbital on the contact carbon atoms @xmath225 , @xmath226 , and @xmath293 , respectively . , and polarization angle @xmath219 .
the chemical potentials are set as @xmath294 , and @xmath295 , resulting in lead 2 and 3 having a reflection symmetry along the x axis.,width=302 ] for a three - terminal system , by virtue of eq .
( [ elec_current02 ] ) , the current in terminal ( lead ) 1 , 2 , and 3 can be expressed as @xmath296,\end{aligned}\ ] ] @xmath297,\end{aligned}\ ] ] @xmath298,\end{aligned}\ ] ] where @xmath299 $ ] for @xmath19 and @xmath300 , @xmath301 , and @xmath302 .
note that the system does not have generalized parity symmetry , i.e. , @xmath303 .
[ fig8 ] shows that @xmath304 and @xmath305 are functions of laser polarization angle @xmath219 , and @xmath304 and @xmath305 are equal at @xmath252 , @xmath306 , and @xmath278 . at @xmath252 , @xmath306 , and @xmath278 ,
lead 2 and 3 have reflection symmetry along the x axis in fig .
however , for any other polarization angle , @xmath304 and @xmath305 differ from each other since the reflection symmetry is destroyed by the laser polarization .
moreover , at @xmath307 , @xmath304 reaches its maximum value while @xmath305 reaches its minimum value .
the maximum value of the current ratio , @xmath308 , is three times higher than the value reported in a four - site model @xcite .
in addition , the maximum current ratio @xmath309 at @xmath310 is also approximately 342 due to the reflection symmetry of lead 2 and 3 .
note that the current ratio depends not only on the symmetry of the router but on laser frequency .
in a three - terminal pam junction for @xmath242 v , @xmath250 ev , @xmath269 v / cm , @xmath311 ev , and @xmath312 ev.,width=302 ] the above results illustrate the correlation between laser polarization and the current ratio , and show that it is possible to manipulate the direction of electric current by using a weak linearly polarized laser field .
a single - molecule router is a new type of device , and its utility deserves further exploration . series and parallel circuit elements are basic electrical network components .
series circuits at the nanoscale , such as electron transport through linear arrays of quantum dots @xcite and alkyl monolayers @xcite , have been extensively studied , but parallel circuits at the nanoscale have not received such attention .
thus , to bolster the foundation of molecular electronics , an understanding of parallel circuits at the molecular level is important . in this section , we investigate electron transport through a parallel circuit based on a simple two - level model in the presence of a high - frequency driving field . , and polarization angle @xmath277 . here ,
@xmath313 and @xmath314 represent the positions of the two dots.,width=302 ] fig .
[ fig9 ] shows a parallel circuit composed of double molecular quantum dots , each of which is connected two leads in the presence of a monochromatic laser field with strength e , frequency @xmath218 , and polarization angle @xmath277 , i.e. , @xmath315 .
the positions of the two molecular quantum dots are @xmath316 and @xmath317 . for simplicity ,
the molecular hamiltonian @xmath1 is described by a tight - binding model within the electric dipole approximation , @xmath318 where @xmath319 corresponds to the electric dipole interaction induced by the laser field , @xmath212 is the hopping integral between the two quantum dots , and @xmath320 , @xmath321 , and @xmath322 are pauli matrices . to facilitate the analysis
, we assume that the on - site energy of the quantum dots is zero . within the wide band limit approximation and the assumption of symmetric molecule - lead coupling ,
the coupling function is @xmath323 in order to reduce the complexity of the problem , we do not consider the off - diagonal terms of the coupling function , which may cause cooperative effects @xcite . by substitution of eq .
( [ hmatrix3_1 ] ) and eq .
( [ coupmatrix3_2 ] ) into eq.([transmission ] ) , the retarded green s function can be divided into four components @xmath324 where @xmath325 due to the system with generalized parity symmetry . using @xmath326 and substituting eq .
( [ 4component1 ] ) into eq .
( [ elec_current02 ] ) , the total current can be decomposed into four current components , @xmath327 where @xmath328 .
\end{aligned}\ ] ] and @xmath329 or @xmath301 .
note that @xmath330 and @xmath331 due to @xmath332 and @xmath326 .
the schematic representation of the four current components is shown in fig .
[ fig10 ] . .
( b ) @xmath333 .
( c ) @xmath334 .
( d ) @xmath335.,width=302 ] v , @xmath336 , @xmath337 ev , @xmath338 , and @xmath339 .
we assume symmetric chemical potentials for the two leads , i.e.,@xmath42 and @xmath43.,width=302 ] fig .
[ fig11 ] shows the current - field intensity characteristics of a parallel circuit based on double molecular quantum dots in a high - frequency driving field ( @xmath338 ) . in the absence of the laser field ,
i.e. , @xmath340 , we can observe @xmath341 and @xmath342 , which can be understood quantitatively by the following discussion .
first , in the absence of the laser field , eq . ( [ current_serial2 ] ) can reduce to @xmath343 . \end{aligned}\ ] ] for a two - level system , by virtue of eq .
( [ green_zerofield2 ] ) , the retarded green s functions @xmath344 and @xmath345 have analytical forms and can be approximated as @xmath346 and @xmath347 the cross terms in eq .
( [ serial_green1 ] ) and eq .
( [ serial_green3 ] ) correspond to constructive and destructive quantum interference between two resonant states , respectively @xcite , and they can be neglected in the resonant tunneling regime .
therefore , we have @xmath348 . in addition , for the condition that the energy levels of all resonant states are between @xmath349 and @xmath350 , we can approximate @xmath351\approx\int^{\infty}_{-\infty}d\epsilon |g^{\textrm{r}(0)}_{11}(\epsilon)|^2 $ ] .
we substitute eqs .
( [ serial_green2 ] ) and ( [ serial_green4 ] ) into eq .
( [ current_serial3 ] ) and make use of @xmath352 , and then we derive @xmath353 and @xmath354 , which are in quantitatively good agreement with the numerical results in fig .
[ fig11 ] . in the presence of the laser field ( @xmath355 ) ,
[ fig11 ] shows that @xmath333 becomes almost zero at @xmath356 , @xmath357 , and @xmath358 while @xmath359 reaches local maximum values at the same positions , indicating that the individual current components can be manipulated by a laser field . to understand this behavior
, we transform the hamiltonian in eq .
( [ hmatrix3_2 ] ) into a rotating frame @xcite @xmath360 by using the operator @xmath361 this transformation gives a good description of the dynamics of a time - periodic system for the high - frequency driving condition @xmath362 and the strong - field driving condition @xmath363 hold @xcite .
using that @xmath364 and @xmath365 , where @xmath366 are bessel functions of the first kind and @xmath367 are integers , we have @xmath368 recall that @xmath369 so we can adopt a high - frequency approximation @xcite and eq .
( [ rotating3 ] ) can be expressed as a static system with the effective hopping integrals @xmath370 as follows @xmath371 eq .
( [ rotating4 ] ) indicates that when @xmath372 , there is no coupling between two molecular quantum dots , leading to @xmath373 .
therefore , we can derive the retarded green s functions in the rotating frame , @xmath374 corresponding to coherent destruction of tunneling @xcite .
in addition , the first three roots of @xmath375 , i.e. , @xmath376 , @xmath377 , and @xmath378 , correspond to @xmath379 , @xmath380 , and @xmath381 , consistent with the numerical results in fig . [ fig11 ] . moreover , no coupling between the two molecular quantum dots means that the on - site wavefunction amplitude @xmath382 or @xmath383 in eq .
( [ green_zerofield2 ] ) , indicating that current only can pass through the dot 1 or 2 .
consequently , @xmath384 reaches local maxima at the first three roots of @xmath375 .
this section showed that the current components of a molecular parallel circuit can be manipulated by a strong high - frequency driving field .
in this study , we develop a new general method to simulate electron transport through a single - molecule junction under time - periodic fields .
this method also enables dealing with electron transport through quantum dots and other nanostructures in a strong driving field . to demonstrate the wide range of applications for this methodology
, we give three examples : single - molecule optoelectronic switches , routers , and parallel circuits , and investigate their transport properties .
our computations show that pam - based optoelectronic switches have robust large on - off switch ratios and weak - field operating conditions , which are not sensitive to asymmetric molecule - lead couplings .
in addition , the magnitude of the current can be tuned by changing the direction of laser polarization and the field amplitude .
the field - amplitude power laws for one- and two - photon assisted tunneling are evident in the computational results , and the laws can be proven by using the perturbation theory . for pam - based optoelectronic routers ,
we show that it is possible to manipulate the direction of electric current through the pam molecule by using a weak linearly polarized laser field . the maximum current ratio @xmath385 depends on the symmetry of the routers , and the ratio can reach approximately 340 . for the parallel circuits made of molecular quantum dots
, the total current can be divided into four components , the magnitude of which can be controlled by a high - frequency laser field .
in addition , we quantitatively derive the value of current in the absence of a laser field and successfully explain the condition of coherent destruction of tunneling by using the rotating wave approximation and the high - frequency approximation .
our study opens up a new direction for exploring light - driven transport through molecular junctions and its potential applications in single - molecule optoelectronics @xcite .
although we have developed a new general method to deal with light - driven transport , several issues remain to be resolved .
first , our floquet analysis is based on the wide band limit approximation , i.e. , no memory effects . for a practical system
, the role of electrodes may be important .
thus , a next step is to develop a more general formulation which enables dealing with the surface green s function of the electrodes .
second , our method is based on a single - electron model , i.e. , an independent electron model , so it does not include many - body effects such as electron - electron interactions and electron - photon interactions @xcite .
lehmann et al .
have investigated vibrational effects in a two - site system @xcite , but they only considered a weak thermal coupling limit , i.e. , no many - body effect such as polarons @xcite in their study .
the synergistic effects of many - body interactions and photon - assisted tunneling may exhibit new physical phenomena calling for study .
third , for simplicity we investigate light - driven transport through the three systems using a hckel - type hamiltonian , but our method can be combined with a more realistic model , e.g. , molecular hamiltonian computed from the density - functional theory in a maximally localized wannier function representation .
high - level simulations may give a fully complete description relevant to experimental investigations of molecular optoelectronics .
fourth , our method can not deal with electronic excitation in a molecular junction .
as the frequencies of light are compatible with the energy difference between the homo ( highest occupied molecular orbital ) and lumo ( lowest unoccupied molecular orbital ) , photon assisted tunneling and electronic excitation may occur simultaneously .
the distinction between the two mechanisms is still an open question .
we hope that this work motivates further theoretical and experimental investigations into light - driven transport through single - molecule junctions .
we thank dr .
tak - san ho for useful discussions .
this research is supported by the nsf ( grant number che-1058644 ) , aro ( grant number w911nf-13 - 1 - 0237 ) and ppst .
substitution of eq .
( [ rate_green2 ] ) into @xmath386 gives @xmath387f_{\alpha}(\epsilon),\end{aligned}\ ] ] in which @xmath388 corresponds to the charge contributed by the lead @xmath19 to the molecule .
it can be proved that @xmath389 since @xmath390 and @xmath391 are periodic functions of time .
integration of the tunneling rate @xmath392 multiplied by the fermi function @xmath393 over a period @xmath46 and use of the fundamental theorem of calculus gives @xmath394 by substitution of eq .
( [ charging2 ] ) into eq .
( [ charging3 ] ) and using the relation @xmath104 , we can obtain @xmath395 indicating that @xmath396 corresponds to periodic charging of the molecule driven by external time - periodic fields and contributes zero current over a period .
in addition , we can invoke eq .
( [ rate_green2 ] ) to prove that @xmath397 is a periodic function of time .
then , it follows that the first two terms in eq .
( [ current1 ] ) average to zero .
eq . ( [ fm ] ) and its adjoint eigenvalue equation can be expressed as @xmath398 where the eigenstates @xmath399 and @xmath400 form a complete biorthogonal basis , i.e. , @xmath401 , and the eigenvalues @xmath402 @xcite . in the previous section , the quasienergies @xmath403 and floquet states @xmath404 are chosen in the first brillouin zone , while here for convenience we choose @xmath403 and @xmath404 which correspond to the energies and the states derived from a hamiltonian without a driving field , i.e. , @xmath405 and @xmath406 in eq .
( [ hamiltonian_zerofield ] ) . as the external laser field is weak
, we can divide the time - averaged floquet hamiltonian into two parts @xmath407 where @xmath408 is a dimensionless parameter ranging continuously from 0 ( zero perturbation ) to 1 ( the full perturbation ) , @xmath409 is a non - hermitian unperturbed hamiltonian , and @xmath410 is a perturbed hamiltonian . according to eq .
( [ supermatrix ] ) , we separate @xmath411 into @xmath412 and @xmath413 we begin with the unperturbed non - hermitian hamiltonian @xmath409 , which has known eigenvalues and eigenstates @xmath414 which satisfy @xmath415 and @xmath416 .
note that @xmath417 and @xmath418 are block diagonal matrices so @xmath419 @xmath420 substituting eqs .
( [ perturbation1 ] ) , ( [ perturbation2_1 ] ) , and ( [ perturbation2_2 ] ) into eq .
( [ perturbation_h1 ] ) , expanding eq .
( [ perturbation_h1 ] ) in powers of @xmath408 , and multiplying by @xmath423 give the first - order correction to the coefficients of the wave function @xmath424_{n''k'',n'k'}\phi^{(0)n',k'}_{\lambda,\zeta}}{q^{(0)}_{\lambda,\zeta}-q^{(0)}_{\lambda',\zeta'}}\phi^{(0)n ,- k}_{\lambda',\zeta'},\end{aligned}\ ] ] where @xmath425 .
( [ perturbation_coeff1 ] ) has a similar form to the first - order correction for the coefficients of the wave function derived from standard time - independent perturbation theory . according to eq .
( [ molhamiltonian ] ) , we obtain @xmath426_{n''k'',n'k ' } \nonumber \\ & = -\frac{\textrm{e}\textrm{e}}{2}(x_{n'}\cos\theta+y_{n'}\sin\theta)\delta_{n'',n'}\delta_{k'',k'\pm 1}.\end{aligned}\ ] ] where the first term is equal to zero and we neglect the fourth term which corresponds to the next - order correction . as a result , we obtain @xmath429 suppose the energy of a tunneling electron is close to the energy of a particular one - photon quasistate @xmath430 , i.e. , @xmath431 , and is far away from the other states . then , eq . ( [ green_pert1 ] ) becomes @xmath432 according to eqs .
( [ perturbation_coeff1 ] ) and ( [ h1_element ] ) , we can show @xmath433 which indicates that the current induced by one - photon assisted tunneling is proportional to @xmath266 .
moreover , by using the second - order correction to the coefficients of the wave function ( not shown here ) , we can derive that @xmath434 and the current induced by two - photon assisted tunneling is proportional to @xmath435 .
note that eq .
( [ perturbation_coeff1 ] ) is generally valid in the case of @xmath436 , e.g. , the high frequency limit .
however , according to eqs .
( [ blockcoef1 ] ) , ( [ blockcoef2 ] ) , and ( [ h1_element ] ) , the conditions @xmath436 can reduce to @xmath437 .
we found that all states satisfy @xmath438 , so it is reasonable to apply the analysis to establish the field - amplitude power laws . | we develop a new general formulation to explore light - driven electron transport through a single - molecule device with multiple pathways .
three individual systems are proposed including ( i ) a two - terminal molecular junction based on phenyl - acetylene macrocycle ( pam ) , ( ii ) pam with three terminals , and ( iii ) a parallel molecular circuit .
the computations show that pam - based optoelectronic switches have robust large on - off ratios and weak - field operating conditions , which are not sensitive to asymmetric molecule - lead couplings .
in addition , field - amplitude power laws for one- and two - photon assisted tunneling are evident in the computational results , and the laws can be proven by using perturbation analysis . for pam - based optoelectronic routers ,
we show that it is possible to manipulate the direction of electric current through the pam molecule by using a weak linearly polarized laser field . for parallel circuits made of molecular quantum dots ,
the condition of coherent destruction of tunneling is derived by using the rotating wave approximation and the high - frequency approximation . |
the qcd - inspired light - front constituent quark model @xcite , with two components in the interaction , a contact term and a coulomb - like potential , was used previously to investigate the properties of mesons @xcite with reasonable success .
it was also applied to investigate the binding energy of the ground state of spin 1/2 @xmath0 baryons @xcite , where the coulomb - like interaction was left out . in that work
, a special regularization scheme was used in which the masses of the virtual two - body subsystems were constrained to be real , as a result the quark binding in the spin 1/2 low - lying states of the @xmath1 , @xmath4 and @xmath3 was qualitatively reproduced .
recently , the integral equation for a three - boson system interacting with pairwise contact interaction in the light - front @xcite was regularized with a sharp cut - off @xcite and applied in the study of the nucleon . in the infinite cut - off limit
it was previously shown that the three - boson system is stable for values of the two - boson bound state mass above a critical value @xcite .
this motivate us to study the effect of different cut - offs in the @xmath0 ground state masses , obtained by solving the light - front integral equations with a contact interaction .
our aim here is to check to which extend the qualitative properties of the quark binding is regularization independent .
we use only the flavor independent contact interaction between the constituent quarks , which brings the physical scale of the ground state of the nucleon and includes the minimal number of physical scales to describe @xmath0 systems . the spin is averaged out .
the model has as inputs the constituent quark masses and the nucleon mass is used to fix the interaction strength . in our previous work @xcite , where the masses of the virtual two - quark subsystems were constrained to be real with the nucleon mass fixed at the experimental value , there was no freedom left in the @xmath0 calculation
. here , even with the nucleon mass kept fixed still the cut - off has some freedom , which will be varied in our calculations . to obtain the @xmath0 masses ,
the mass of one of the constituent quark @xmath5 is varied while the strength of the effective contact interaction is supposed flavor independent .
this work is organized as follows . in sec.2
, we briefly discuss the coupled integral equations for the @xmath0 bound - state in the light - front and the regularization schemes . in sec . 3 , we present the numerical results for the binding energies of the @xmath6 , @xmath4 and @xmath7 and give a summary of the work with our conclusion .
with the assumption that the spin is averaged out , the @xmath0 system interacting through a contact pairwise force is represented by two spectator functions or faddeev components of the vertex , which satisfies the coupled bethe - salpeter type equations , derived elsewhere @xcite , and shown diagrammatically in fig . 1 . system . ,
title="fig:",width=264 ] system .
, title="fig:",width=340 ] the coupled integral equations for the fadeev components of the vertex function of the @xmath0 system , represented in fig . 1 , are given by @xcite : @xmath8 @xmath9 , \label{e26}\end{aligned}\ ] ] where @xmath10 and @xmath11 are the @xmath0 mass and four - momentum , respectively .
the masses of the virtual two - quark subsystems are @xmath12 and @xmath13 , and in the baryon rest frame are given by : @xmath14 with ( @xmath15,@xmath16)=@xmath17 or @xmath18 . in our previous work @xcite , we constrained the virtual two - body subsystem masses to real values and using eq .
( [ mass3a ] ) with @xmath19 and @xmath20 substituted by @xmath21 and @xmath22 , respectively , giving the lower bounds for the integration in @xmath22 and the maximum value for the transverse momentum , @xmath23 , for each vertex depending on @xmath22 , as well . in that case
, the theta functions bring to eqs .
( [ mass3 ] ) and ( [ e26 ] ) the discussed constraint .
the two - quark scattering amplitudes , @xmath24 , are the solutions of the bethe - salpeter equations in the ladder approximation for a contact interaction between the quarks . as a function of the baryon mass @xmath10 . attributed experimental binding energies for the spin 1/2 low - lying states of the nucleon , @xmath1 , @xmath2 and @xmath3 from @xcite ( full squares ) .
model results from ref .
@xcite ( solid curve ) .
present calculations with transverse momentum cut - off with the values of @xmath25 ( dashed curve ) , @xmath26 ( dotted curve ) and @xmath27 ( dot - dashed curve).,width=321 ] here , we use a different regularization scheme of eqs .
( [ mass3 ] ) and ( [ e26 ] ) , where @xmath28 and the theta functions in @xmath22 are dropped . the nucleon mass is fixed to its experimental value , therefore we have to adjust the renormalized coupling constant for each value of the transverse momentum cut - off @xmath29^{-1}\ , \label{tau}\end{aligned}\ ] ] where @xmath30 is the bare interaction strength . as its stands , eq .
( [ tau ] ) is ill - defined .
the renormalization condition is chosen at an arbitrary subtraction mass point @xmath31 where the value of @xmath32 is known and given by @xmath33 , the renormalized interaction strength , which is enough to remove the logarithmic divergence in eq .
( [ tau ] ) .
therefore , the bare strength is given as : @xmath34 which is supposed to be flavor independent . the strength of the effective interaction is determined by fitting the nucleon mass .
the physical inputs of the model defined by eqs .
( [ mass3 ] ) and ( [ e26 ] ) , are the renormalized interaction strength @xmath35 , the constituent quark masses @xmath36 gev and @xmath37 ( @xmath38 ) which were found in @xcite , where as well , it was attributed a binding energy to the spin 1/2 baryons ( nucleon , @xmath1 , @xmath4 and @xmath39 defined as @xmath40 . considering that the flavor - off - diagonal vector mesons are weakly bound systems of constituent quarks in the qcd - inspired model of ref .
@xcite , we obtained values of the constituent quark masses .
the attributed values to the binding energies were found @xcite using constituent quark masses and the experimental baryon masses from @xcite .
the attributed values for the constituent quark binding energies in the spin 1/2 low - lying states of the nucleon , @xmath41 , @xmath4 and @xmath3 are shown in fig . 2 by the full squares , from the left to the right , respectively .
in fig . 2 , we show results for the calculation of the binding energy of the baryon for different renormalization schemes and values of the transverse momentum cut - off compared to the attributed baryon binding energies . for the values of the cut - off , @xmath25 , @xmath26 and @xmath27
, we fit the renormalized strength of the interaction to reproduce the nucleon mass . by changing the mass of the heavy quark ,
while the strength of the interaction was kept unchanged , the coupled integral equations were solved to obtain the mass and binding energy of the baryons .
again , as observed in the work @xcite for the baryon mass above @xmath42 gev , the bound @xmath0 system goes to the diquark threshold .
our results show that the general behavior of the experimental data of masses and binding energies of the spin 1/2 low - lying states of @xmath43 , @xmath4 and @xmath3 are reproduced with different momentum cut - off s constrained by the value of the nucleon mass . in summary , in the present contribution
we have studied how different regularization schemes in the relativistic coupled @xmath0 homogeneous integral equation affects the qualitative properties of the quark binding in @xmath0 baryons .
our calculations indicate that the general features of the low - lying spin 1/2 baryon binding energy as a function of its mass , is model independent to a large extent ( small cut - off sensitivity ) as long as the relativistic @xmath0 model has a flavor independent effective short - range interaction . therefore , the conclusions drawn in our previous work @xcite does not qualitatively depend on the regularization scheme .
we would like to thank the brazilian funding agencies fapesp ( fundao de amparo a pesquisa do estado de so paulo ) and cnpq ( conselho nacional de desenvolvimento cientfico e tecnolgico ) .
efs thanks the organizing committee of the conference for the kindly invitation to present this work .
99 h .- c .
pauli , eur .
j. * c7 * , ( 1998 ) 289 ; h .- c .
pauli , in : new directions in quantum chromodynamics , c.r.ji and d.p .
min , eds . , aip(1999 ) 80 - 139 ; nucl .
* b * ( proc . suppl . )
* 90 * ( 2000 ) 154 ; ibid . , ( 2000 ) 259 .
brodsky , h .- c .
pauli , and s. s. pinsky , phys .
* 301 * ( 1998 ) 299 .
t. frederico , h .- c .
pauli , phys . rev .
* d64 * ( 2001 ) 054007 .
suisso , j.p.b.c . de melo , and t. frederico , phys . rev .
* d65 * ( 2002 ) 094009 .
t. frederico , phys.lett . *
b282 * ( 1992 ) 409 ; s.k .
adhikari , l. tomio and t. frederico , ann . phys .
* 235*(1994 ) 77 . m. beyer , s. mattiello , t. frederico , h.j .
weber , to be published in few - body systems , nucl - th/0302018 .
j. carbonell , v.a .
karmanov , phys . rev .
* c67 * ( 2003 ) 037001 .
k. hagiwara et al .
, phys . rev .
* d66 * ( 2002 ) 010001 . | we study the mass of the ground state of @xmath0 systems using different regularization schemes of the relativistic integral equation obtained with a flavor independent contact interaction in a qcd - inspired light - front model .
we calculate the masses of the spin 1/2 low - lying states of the @xmath1 , @xmath2 and @xmath3 for different values of the regularization cut - off parameter with a fixed nucleon mass .
our results are in remarkable agreement with the experimental data . |
in recent years many experimental and theoretical efforts@xmath8 have been made to investigate on the nature of the bremsstrahlung emission in the @xmath0-decay of heavy nuclei , because the behavior of the energy spectrum of photons is strongly related to the dynamics of the @xmath0-decay and alpha nucleus particle potential . in some case
the energy spectrum of bremsstrahlung shows some slight oscillations@xcite , in other case authors observed a minimum@xcite , in some experiment authors@xcite have not observed evidence of any structure .
therefore the main problem to investigate the tunneling of the @xmath0-particle through the coulomb barrier of nucleus is to enlarge the area of the study on other nuclei and to compare the characteristics of the @xmath3-spectra . in this paper
we present the results of the last experiment on the study of the bremsstrahlung emission in @xmath0-decay of @xmath1 .
we also analyze and discuss the comparison between the experimental and theoretical results of the photon emission related to the @xmath1 and @xmath5 nuclei . in sec .
2 we shortly describe the characteristics of the experiment and the results of the bremsstrahlung emission probability . in sec .
[ sec.3 ] we present the basis of the model , the analysis of our calculation about the various contributions of the bremsstrahlung emission , and we discuss the results of the @xmath3-emission for the @xmath0-decay of @xmath1 , in comparison with the ones previously obtained for @xmath5 . sec .
4 is devoted to our conclusion .
the experimental set up was the same as described in our previous paper@xcite .
the source of @xmath1 with a activity of about @xmath9 @xmath0-particles@xmath10s was used . along the decay chain of this nucleus
the @xmath0-particles were recorded .
the diameter of the radioactive spot on the source surface was about 8 mm and @xmath0-particles were detected by a silicon surface - barrier detector with energy resolution of about 20 kev at the @xmath0-particle energy of 4.8 mev .
the @xmath0-detector was 200 mm@xmath11 in area and was placed at the distance of about 1 cm from the source .
the time resolution of the @xmath0-@xmath3 coincidence technique was @xmath12= 10 ns .
the @xmath3-rays were detected by the nai(tl)-detector with diameter of 3 cm and thickness of 3 cm , and the distance between source and @xmath3-detector has been about 1.6 cm .
the angles between the two detectors and the normal axis to the surface of the source were @xmath13 .
so the total angle between the @xmath0 and @xmath3 detectors was about @xmath14 .
this angle value was chosen to increase the yield of e1 dipole bremsstrahlung photons and to reduce the influence of @xmath3-rays from e2-quadrupole transitions of excited states of daughter nuclei .
the absolute values of the nai(tl)-detector efficiency was determined by measuring the intensities of lines of the standard @xmath3-sources @xmath15 ( e@xmath6=59.6 kev ) , @xmath16 ( 122 and 136 kev ) , @xmath1 ( 186 , 295 , 532 and 609.4 kev ) , @xmath17 ( 662 kev ) and @xmath18 ( 1.17 and 1.33 mev ) , by replacing each of them in the position of the @xmath1-source of @xmath0-particles.the measurements of bremsstrahlung photons in coincidence with @xmath0-particles for five @xmath0-groups@xcite from decays of @xmath1 ( e@xmath2= 4.784 mev ) , @xmath19 ( 5.304 mev ) , @xmath20 ( 5.490 mev ) , @xmath21 ( 6.002 mev ) and @xmath5 ( 7.687 mev ) were performed during about 1500 hours and the total number of 4.4@xmath2210@xmath23 events was registered by the @xmath0-@xmath3 coincidence system .
the analysis of these events in the ( e@xmath6 vs e@xmath2)-plane located in the region of the total energy conservation line e@xmath6 + e@xmath2= _ constant _ , taking into account the detectors energy resolutions ( 20 kev for @xmath0-particles and 32 kev for photons ) , gives us the possibility to determine the yield of photons at energies up to 803 kev ( 1-st excited state of the daughter @xmath24pb nucleus ) for the @xmath0-decay of @xmath25 .
the range of ( e@xmath6 vs e@xmath2)-coincidences was collected for one of the experimental run with the measurement time of about 150 hours . for other details on the recorded events and analysis ,
see our previous paper @xcite and references therein . to take into account the defined angular dimensions of detectors we use the same procedure to obtain the angular averaged probability of the photon emission dp / de@xmath6 as described in refs . , where the @xmath26 angular correlation function w(@xmath27 ) , for the case of e1-dipole photon emission and point - like @xmath26 source is @xmath28 where a@xmath29= -1 for the dipole e1 transitions , p@xmath29(cos@xmath27 ) is the second order polynome of legendre , q@xmath29 is the total geometrical attenuation coefficient of the second order : q@xmath29= q@xmath30 @xmath22 q@xmath31 , q@xmath32 being the geometrical attenuation coefficients for @xmath0-particle and photon detectors . the exact calculations in the framework of the density matrix theory@xcite give us the real experimental geometry values of the attenuation coefficients : for the @xmath0-particle detector it is q@xmath30=0.90 , and for the photon detector the value of q@xmath31-coefficent is varied from q@xmath31= 0.66 for e@xmath6= 100 kev up to q@xmath31= 0.76 for e@xmath6= 800 kev .
the probability of the photon emission at angle @xmath27 can be written as@xcite @xmath33 where @xmath34 is the total number of @xmath26 coincidences during the measurement time @xmath35 , in the intervals of photon energies @xmath36 and angles @xmath37 , @xmath38 is the intensity of particles in the @xmath0-detector and @xmath39 is the absolute efficiency of the @xmath3-detector .
therefore , the total probability of the bremsstrahlung emission is @xmath40 the check on the experimental data has been made by measuring the coincidences between the @xmath0-particles emitted to the first excited level of the daughter nuclei @xmath41rn , @xmath21 and @xmath42 , corresponding to @xmath3-rays with energies 186 , 510 and 800 kev@xcite .
the angular correlation function w(@xmath27 ) , in the case of @xmath3-ray e2-transitions , can be presented as@xcite @xmath43 where @xmath44 = 5/7 , @xmath45= -12/7 , @xmath46 = @xmath47 , @xmath48 = 0.68 , @xmath49 = 0.16 for @xmath50 = 100 kev and @xmath49 = 0.37 for @xmath50 = 800 kev . the bremsstrahlung spectra have been averaged over a photon energy interval of 25 kev .
the measured values of the photon emission probability dp / de@xmath6 due to the bremsstrahlung process accompanying the @xmath0-decay of @xmath1 are shown by solid squares in fig .
[ fig:226ra_1 ] . in the following sec.[sec.3 ] we calculate the bremsstrahlung spectrum accompanying the @xmath0-decay of @xmath1 and present , for a comparison , the above results for @xmath1 ( where is e@xmath2=4.8 mev ) with the ones presented in our previous paper @xcite for @xmath5 ( where is e@xmath2=7.7 mev ) . in the theoretical description of the bremsstrahlung emission during the @xmath0-decay
there are some problems : one is the choice of the realistic wave function of the @xmath0-particle inside the nuclear potential@xcite .
for example , the shape of the nuclear potential ( the values of the nuclear radius r@xmath51 and the deepness v@xmath51 for a rectangular potential ) influences the slope of the wave function near the nuclear surface and therefore the conditions of tunneling through the coulomb barrier .
other problems can be connected with the influence of the nuclear surface deformation and electron screening of the coulomb barrier .
we define the bremsstrahlung probability during the @xmath0-decay of a nucleus in terms of the transition matrix elements for the compound quantum system ( @xmath0-particle and daughter nucleus ) from its state before photon emission ( we name such a state as the _ initial @xmath52-state _ ) into its state after the photon emission ( we name such a state as the _ final @xmath53-state _ ) .
if it is possible to separate total wave function of @xmath0-particle ( before and after photon emission ) into radial and spherical symmetric components ( as in the approximation of the spherically symmetric @xmath0-decay ) , then one can find the expression for the total bremsstrahlung probability with the separation on the radial and angular components explicitly by an analytical way .
here , the radius defines the position of the particle with reduced mass relatively to the center of mass .
the angular components contain all the detailed information about the directions of this particle motion ( with taking into account its tunneling ) before and after the photon emission and on the direction of the photon emission . according to ref .
we define the bremsstrahlung probability as @xmath54 where @xmath55 @xmath56 @xmath57 in ( [ eq.3.1.3 ] ) @xmath58 is the radial integral independent on the angle @xmath59 : @xmath60 in determination of the wave functions of the initial and final states we use the selection rules for the quantum numbers @xmath61 and @xmath62 : @xmath63 in ( [ eq.3.1.5 ] ) @xmath64 is the spherical bessel function of the order @xmath61 , @xmath65 and @xmath66 are the radial components of the total wave functions @xmath67 and @xmath68 of the system in the initial @xmath52- and final @xmath53-state , respectively . for other notations and details
@xcite , in accordance with refs . .
to describe the interaction between the @xmath0-particle and daughter nucleus ( @xmath69 , @xmath70 ) we use the following potential @xcite : @xmath71 where coulomb @xmath72 , nuclear @xmath73 and centrifugal @xmath74 components have such form : @xmath75 and @xmath76 here , @xmath77 is the @xmath77-value for the @xmath0-decay , @xmath78 is the radius of the daughter nucleus , @xmath79 is the strength of the nuclear component ; @xmath80 is the effective radius of the nuclear component , @xmath81 is the parameter of the diffuseness ; @xmath82 is the spherical harmonic function of the second order , @xmath27 is the angle between the direction of the leaving @xmath0-particle and the axis of the axial symmetry of the daughter nucleus ; @xmath83 is the parameter of the quadruple deformation of the daughter nucleus .
the parameters of the coulomb and nuclear components are defined in refs . .
in order to obtain the spectrum , we have to know wave functions in the initial and final states . in the spherically symmetric approximation one can rewrite the total wave functions by separating the radial and angular components : @xmath84 we find the radial components @xmath85 numerically on the base of the given alpha - nucleus potential . here
, we use the following boundary conditions : the @xmath52-state of the system before the photon emission is a pure decaying state , and therefore for its description we use the wave function for the @xmath0-decay ; after the photon emission the state of the system is changed and it is more convenient to use the wave function as the scattering of the @xmath0-particle by the daughter nucleus for the description of the @xmath53-state .
so , we impose the following boundary conditions on the radial components @xmath85 : @xmath86 where @xmath87 and @xmath88 are the coulomb functions as used in refs . .
it is interesting to estimate how much is the emission of photons from the tunneling region and the ones from the internal and external regions concerning the barrier versus the distance @xmath89 .
such a question was put for the first time by j. kasagi _ et al . _ in ref . in the tentative to explain the difference between the experimental data and the calculated spectrum for the @xmath90 nucleus .
some later , a constructive analysis was proposed by n. takigawa _ et al . _ in ref . and , independently e. tkalya in ref . , both giving the theoretical basis to work with the contributions of the photon emission from different regions and interference term by different approaches .
as we see and shall show further , the study of the photon emission from the different spacial regions allows one not only to analyze the difference between the experimental data and the calculated spectrum for @xmath90 , but gives a more important information about the studied process . for this reason
, we shall develop a formalism of the photon emission from the different spacial regions in the framework of our model .
let us define the bremsstrahlung probabilities of photons emitted from different spacial regions , characterized by the points @xmath91 and @xmath92 in fig .
[ barrier ] . according to ref .
, we shall assume that the emission of photons from the internal spacial region ( defined from @xmath93 up to @xmath94 ) is very small in a comparison with the total photon emission , and therefore we can neglect it . in practical calculations of the spectra
, we find the integral ( [ eq.3.1.5 ] ) by a reasonable approximation taking an any finite value @xmath95 into account instead of @xmath96 . by such reasons
, we shall analyze the integral ( [ eq.3.1.5 ] ) only inside the spacial intervals from the point @xmath97 up to @xmath95 .
we separate this integral into two items : @xmath98 where @xmath99 where ( @xmath100 ) is the tunneling region , ( @xmath101 ) is the external one , and @xmath102 is the energy of the system in the initial @xmath52-state .
then , the formula ( [ eq.3.1.1 ] ) of the bremsstrahlung probability is transformed into the following : @xmath103 where the three components @xmath104 , @xmath105 and @xmath106 are defined through the finite integrals @xmath107 and @xmath108 . in particular , at @xmath109 ( and assuming that the radial wave function @xmath66 in the final @xmath53-state
does not depend on quantum number @xmath110 at @xmath111 like as coulomb functions ) one can obtain : @xmath112 @xmath113 @xmath114 therefore , we affirm the following : the total bremsstrahlung spectrum is not simply the summation of the direct ( pure ) probabilities from tunneling and external regions @xmath104 and @xmath105 , but it also includes the interference term @xmath106 ( see also refs . ) ; the probabilities @xmath104 and @xmath105 from tunneling and external regions are only positive , the interference term @xmath106 can be positive or negative .
we have applied the above described method to calculate the bremsstrahlung spectrum emitted during the @xmath0-decay of the @xmath115 nucleus .
the results are presented by the full line in fig .
[ fig:226ra_1 ] together with the experimental data ( solid squares ) . in calculation
we have used the approximation of the spherically symmetric @xmath0-decay @xmath109 only in determination of @xmath116 in ( [ eq.3.1.3 ] ) , and the angle @xmath117 between the directions of the @xmath0-particle motion ( with possible tunneling ) and the photon emission .
the energy @xmath118 of the @xmath0-particle in the initial @xmath52-state is @xmath119 mev , according to ref . .
the approximation of the spherically symmetric @xmath0-decay is used in the present paper for the following reasons ( see also the analysis discussed in ref . ) . at the range of the nucleus surface and maximum value of the coulomb barrier
, the potential form changes more rapidly than for larger distances outside the nucleus surface and also inside the nucleus ( see fig . [ barrier ] ) .
therefore at larger distances from the nucleus ( and also inside nucleus ) , where the changing of the potential is much slower than in the surface region , the emission photon energy and influence of the deformation parameter of the nucleus both are smaller than near the surface region .
considering that the external region ( @xmath101 ) included between several fm and more large distances ( much and further ) is very wide , the photon emission probability from such wide space is bigger than the one from the narrow range of the nuclear surface region ( @xmath100 ) . therefore , the description of the bremsstrahlung spectrum of fig .
[ fig:226ra_1 ] by the spherical - nucleus approximation is rather good for lower photon energies ( near about 100300 kev where the @xmath3-emission probability @xmath120 is high ) .
of course , for photon energies larger than 400 kev it is useful to take into account the deformation parameter of the @xmath1 nucleus ( @xmath121=0.151)@xcite . such an improvement of the theory that takes into account the deformation parameter of the nucleus will be made in the next future because the upgrading of the model overcomes the aim of the present paper . in fig .
[ fig:226ra_214po_full ] we report , for a comparison , the presented results for @xmath1 together with the ones found for the @xmath5 nucleus@xcite . as fig .
[ fig:226ra_214po_full ] shows , both experimental and theoretical results of the photon emission probability obtained for the @xmath0-decay of @xmath1 are clearly lower than the ones obtained for @xmath5 .
the difference between the two sets of data can be attributed to the different structure of the two nuclei , which affects the motion of the @xmath0-particle inside the barrier .
the ratio between the two sets of data of the photon emission probability @xmath120 is strongly characterized by the different @xmath0-decay energy for @xmath5 ( e@xmath2=7.7 mev ) and @xmath1 ( e@xmath2=4.8 mev ) concerning the shapes of the alpha - nucleus barriers for these nuclei . in the @xmath0-decay of @xmath5 , the @xmath0-particle with energy of 7.7
mev passes under the barrier in the upper part where the potential form changes more strongly emitting photons of high energies , while in the case of @xmath1 the @xmath0-particle with energy of 4.8 mev passes under the barrier in the lower part where the potential changes partially more slowly emitting photons of lower energies .
to explain the difference between the slopes of the bremsstrahlung spectra for the two considered nuclei , we formulate the following consideration : the slope of the bremsstrahlung spectrum is defined directly by the principal difference between the emission of photons during tunneling of the @xmath0-particle and the emission of photons during its motion . to understand
why two nuclei have such different slopes of the bremsstrahlung spectra , we estimate how much the emission of photons during tunneling of the @xmath0-particle differs from the emission of photons during further motion of this @xmath0-particle in the external region . in the fig .
[ fig:226ra_214po ] we report the tunneling and external contributions of the @xmath3-emission accompanying the @xmath0-decay of @xmath5 and @xmath1 nuclei and the interference terms which are calculated by ( [ eq.3.4.4])([eq.3.4.6 ] ) . to make the analysis clearer , here we present the different contributions and interference terms for each nucleus separately . from calculation represented in this figure
we establish that : for both nuclei , the interference term has negative values in the whole energy region of the emitted photons and plays a destructive role in addition to the contributions from tunneling and external regions in the forming total spectrum ; the calculated total emission probability ( full line ) for @xmath1 is always smaller than the one for @xmath5 , and the external emission probability ( dash - dotted line ) for @xmath1 is smaller than the one for @xmath5 , for photon energies about e@xmath122 120 kev . since the @xmath0-particle energies for the decaying @xmath5 and @xmath1 nuclei are 7.7 mev and 4.8 mev , respectively , the tunneling region ( @xmath123 ) for @xmath1 is longer than for the case of @xmath5 ( see fig .
[ barrier ] ) .
therefore , the bremsstrahlung photons are emitting through a greater distance under the barrier of the @xmath1 nucleus than for @xmath5 .
for this reason , the relative contribution of the photon emission from the tunneling region into the total spectrum for @xmath1 is larger than for @xmath5 , at least up to about e@xmath6=450 kev ( see the dashed lines in panels a ) and b ) of fig .
[ fig:226ra_214po ] ) . in our results ,
the relative contribution of the photon emission from the external region ( @xmath124 ) into the total spectrum for @xmath1 is smaller than for @xmath5 ( see dash - dotted lines in the cited panels a ) and b ) ) , because ( @xmath124 ) is more large for @xmath5 ( see fig .
[ barrier ] ) , but such a contribution from the external region overcomes the total bremsstrahlung spectrum and experimental data for both cases of the @xmath1 and @xmath5 nuclei .
this result clearly confirms that without the appreciable contribution of the photon emission during tunneling and the interference term ( negative for these cases ) , it is impossible that the total bremsstrahlung emission can reach and agree with the experimental results . if the photon emission during tunneling of @xmath0-particle is smaller than the photon emission during the external motion of this @xmath0-particle , then we have to obtain smaller total bremsstrahlung spectrum for @xmath1 than the spectrum for @xmath5 ( at all photon energies ) . figures [ fig:226ra_214po_full ] and [ fig:226ra_214po ] show this effect and confirm ( theoretically and experimentally ) our results and considerations .
the smaller values of the calculated total emission probability for @xmath1 than the one for @xmath5 ( and the smaller values of the external emission probability for @xmath1 than the one for @xmath5 can be explained by a consequence of the fact that outside the barrier the coulomb field ( and its derivative respect to @xmath89 ) , that acts on the @xmath0-particle , in the case of @xmath1 is smaller than in the case of @xmath5 because the external wide region results for the @xmath5 nucleus larger than for @xmath1 and therefore the @xmath3-emission probability for the @xmath5 nucleus is bigger .
we have obtained a good agreement between theory and experiment for the bremsstrahlung spectrum of photons emitted during the @xmath0-decay of the @xmath1 nucleus , for e@xmath6 energies up to about 500 kev .
we think that for photon energies higher than 400 kev it is useful to take into account the deformation parameter of the decaying nucleus in the model , in order to reach a complete good agreement between data and calculation also at higher energies .
we affirm that the photons with higher energies are emitted near the narrow region of the barrier where the potential form changes more rapidly , while the photons of lower energies are emitted at larger distances form the nucleus and in a very wide space where the changing of the potential is much slower .
therefore such a large wide space contributes with photons of low energies but with a higher emission probability .
moreover , we find that the slope of the bremsstrahlung spectrum accompanying the @xmath0-decay of the @xmath1 nucleus decreases more swiftly than the one registered for @xmath5 because the @xmath0-particle energy is lower ( e@xmath2 = 4.8 mev ) for @xmath1 than for @xmath5 ( e@xmath2 = 7.7 mev ) , and this plays an important role both on the motion of the @xmath0-particle and in the determination of the photon emission probability in tunneling and external regions . in both cases we find a destructive interference between the photons emitted from the two above - mentioned regions , and , for the first time , we find an important contribution of the bremsstrahlung photons emission during tunneling of the @xmath0-particle in @xmath7ra , taking into account the realistic @xmath0-nucleus potential in the model .
g. giardina is grateful to the fondazione bonino - pulejo ( fbp ) of messina for the important support received in the international collaboration between the messina group and the moscow state university ( russia ) with the institute for nuclear research of kiev ( ukraine ) . v. s. olkhovsky and s. p.
maydanyuk thank the dipartimento di fisica delluniversit di messina for warm hospitality .
g. giardina , g. fazio , g. mandaglio , m. manganaro , c. sacc , n.v .
eremin , a.a .
paskhalov , d.a .
smirnov , s.p .
maydanyuk and v.s .
olkhovsky , in press in eur .
j. * a 36 * , ( 2008)(published on - line doi 10.1140/epja / i2007 - 10548 - 5 ) | we obtained the spectrum of probability of the bremsstrahlung emission accompanying the @xmath0-decay of @xmath1 ( e@xmath2=4.8 mev ) by measuring the @xmath0-@xmath3 coincidences and using the model presented in our previous study on the @xmath4decay of @xmath5 ( e@xmath2=7.7 mev ) .
we compare the experimental data with the quantum mechanical calculation and find a good agreement between theory and experiment .
we discuss the differences between the photon spectra connected with the @xmath0-decay of the @xmath1 and @xmath5 nuclei .
for the two mentioned nuclei we analyze the bremsstrahlung emission contributions from the tunneling and external regions of the nucleus barrier into the total spectrum , and we find the destructive interference between these contributions .
we also find that the emission of photons during tunneling of the @xmath0-particle gives an important contribution to the bremsstrahlung spectrum in the whole e@xmath6 energy range of the studied @xmath7ra nucleus . |
magnetars are neutron stars with typically longer periods of spinning ( @xmath22@xmath312 s ) and stronger surface magnetic fields ( @xmath0@xmath4 g ) @xcite , compared to normal neutron stars .
their manifestations include soft gamma - ray repeaters ( sgrs ) and anomalous x - ray pulsars ( axps ) , which are mostly located at low galactic latitudes @xcite .
magnetars emit mainly by releasing magnetic energy rather than rotational energy @xcite . due to unusually high surface magnetic fields ,
frequent starquakes occur on magnetars @xcite , leading to recurrent x - ray outbursts of magnetars ( cf .
* and references therein gcn circulars : http://gcn.gsfc.nasa.gov/gcn3_archive.html ] ) .
whenever a magnetar outburst takes place , a substantial amount of energy spills , following the ejecta , to the surroundings and can affect the emission in that region at energies above 1 kev .
whether the gigaelectronvolt ( gev)-emissions from regions of magnetars react correspondingly to magnetar outbursts or not deserves to be a focus of our attention .
pulsed gev radiation from the magnetospheres of magnetars have been predicted by theoretical models ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
nonetheless , it is arguable that the gev pulsations of magnetars are likely to be buried due to their low flux levels ( cf . * ? ? ?
the difficulties in pulsation search of magnetars in @xmath5-ray are further increased by the complicated @xmath5-ray backgrounds in magnetars neighbourhoods , because the galactic diffuse emission is strong and the distributions of celestial objects are crowded at galactic latitudes within @xmath61@xmath7 .
celestial objects in magnetars neighbourhoods include , but are not limited to , magnetars derivative products . while analyses of _ fermi _ large area telescope ( lat ) data did not give convincing evidence for @xmath5-rays from any magnetar ( cf .
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , it is interesting to observe @xmath5-rays from some derivative products of magnetars , as some of them are also magnetically powered and enable us to study physics of ultra - high magnetic field .
supernova remnants ( snrs ) are common derivative products of magnetars as well as other categories of neutron stars , and form through stellar collapse and subsequent supernova explosion .
thanks to lat observations , plenty of gev - bright snr@xmath3molecular cloud ( mc ) association systems have been significantly detected ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and discovered to have broken - power - law or log - parabola spectra @xcite , suggesting an escape of over - energetic particles from the acceleration sites .
pulsar wind nebulae ( pwne ) are another type of pulsars derivative products , which form through pulsars wind materials interacting with and shocked by ambient medium .
they have been studied from the radio band ( synchrotron radiation ; e.g. , * ? ? ?
* ) to the @xmath5-ray band .
at least 27 pulsars have their pwne identified as teraelectronvolt ( tev ) sources ( cf . * ? ? ?
* ) emitting through inverse - compton ( ic ) scattering ( reviewed by * ? ? ? * ) .
until now , swift j1834.9 - 0846 is the only magnetar confirmed to produce an x - ray pwn @xcite , and cxou j171405.7 - 381031 is the only magnetar argued as producing a tev pwn @xcite .
1e 1841 - 045 is an axp discovered to be the central source of snr kes 73 @xcite .
the kes 73/1e 1841 - 045 system has a distance of @xmath8 kpc from us and an age of 750 - 2100 yr @xcite .
this system is potentially responsible for particle acceleration , as it shows intense tev emission at the northern edge of hess j1841 - 055 @xcite .
this arouses our motivation to search for the gev counterpart of the magnetar / snr pair .
1e 1841 - 045 has a spin - down power of @xmath2@xmath9 erg s@xmath10 and a surface magnetic field of @xmath2@xmath11 g @xcite .
1e 1841 - 045 experienced a series of 10 x - ray outbursts in 2011 february@xmath3july @xcite , and it is interesting to see how the gev flux varies shortly after these outbursts .
tight interaction of kes 73 with mcs is supported by detection of oh ( 1720 mhz ) maser emission @xcite , discovery of the methanol source iras 18379 - 0500 which is associated with an h ii region @xcite , and revelation of broad - molecular - line ( bml ) regions by h i and co observations @xcite .
nevertheless , the total solid angle subtended at kes 73 by its adjacent compact h ii regions is only @xmath25% of an entire sphere ( cf .
figure 1(a ) of * ? ? ?
* ) and , to some degree , limits the @xmath5-ray contribution from kes 73 . in this paper , we explore the hundred - megaelectronvolt - to - gev emission in the field of the kes 73/1e 1841 - 045 system by using @xmath27.5 years of _ fermi _ lat data with the latest instrumental responses and background models .
then , we provide some insight into the possible origin(s ) of the @xmath5-rays , in terms of the spectral shapes , energy budgets and long - term temporal behaviors . results of a nearby region consisting of 3xmm j185246.6 + 003317 will be published in a follow - up paper .
we performed a series of binned maximum - likelihood analyses for a 20@xmath7@xmath1220@xmath7 region - of - interest ( roi ) centered at ra=@xmath13 , dec=@xmath14 ( j2000 ) , which is approximately the midpoint between 1e 1841 - 045 and 3xmm j185246.6 + 003317 .
we used the data obtained by lat between 2008 august 4 and 2016 february 1 .
the data were reduced and analyzed with the aid of _ fermi _ science tools v10r0p5 package . in view of the complicated environment of the galactic plane regions
, we adopted the events classified as pass8 clean " class for the analysis so as to better suppress the background . the corresponding instrument response function ( irf )
p8r2@xmath15clean@xmath15v6 " is used throughout the investigation .
we further filtered the data by accepting only the good time intervals where the roi was observed at a zenith angle less than 90@xmath7 so as to reduce the contamination from the albedo of earth . for subtracting the background contribution
, we included the galactic diffuse background ( gll@[email protected] ) , the isotropic background ( iso@xmath15p8r2@xmath15clean@xmath15v6@[email protected] for psf3 " data , iso@xmath15p8r2@xmath15clean@xmath15v6@[email protected] for front " data or iso@xmath15p8r2@xmath15clean@[email protected] for a full set of data ) as well as all other point sources cataloged in the most updated fermi / lat catalog ( 3fgl ; * ? ? ?
* ) within 25@xmath7 from the roi center in the source model . in [
sec : spatial ] , we report discoveries of a few sources within 5@xmath7 from the roi center , which were also included in the source model .
we set free the spectral parameters of the sources within 8@xmath7 from the roi center in the analysis .
for the sources beyond 8@xmath7 from the roi center , their spectral parameters were fixed at the catalog values . in spectral and temporal analysis
, we required each energy - bin and time - segment to attain a signal - to - noise ratio @xmath16@xmath17 ( equivalently , a ts value @xmath169 and a chance probability @xmath180.3% ) for a robust result . for each energy - bin or time - segment _ dissatisfying _ this requirement , we placed a 3.0@xmath19 upper limit on its flux .
we investigate the morphology of the kes 73/1e 1841 - 045 region in high - energy ( he ; @xmath01 gev ) and very - high - energy ( vhe ; @xmath010 gev ) regimes respectively .
we notice that the psf at 10 gev for a full set of data is smaller than that at 1 gev for psf3 " data ( cf .
slac fermi lat performance : http://www.slac.stanford.edu/exp/glast/groups/canda/lat_performance.htm ] ) .
therefore , in order to achieve a compromise between good spatial resolution and adequate photon statistics , we independently analyze two sets of data for morphological studies : ( 1 ) psf3 " data in 1@xmath350 gev , ( 2 ) and a full set of data in 10@xmath3200 gev .
the test - statistic ( ts ) maps of the field around 1e 1841 - 045 are shown in figure [ 1e_tsmap ] , where all 3fgl catalog sources including hess j1841 - 055 are subtracted .
the peak detection significance is @xmath2@xmath20 in 1@xmath350 gev and @xmath2@xmath21 in 10@xmath3200 gev . on the ts map in 1@xmath350 gev , we determined the contour where the ts value is lower than the maximum by 9.488 .
since this contour is very ellipse - like , we fit an ellipse to it .
the center of this ellipse is determined to be the best - fit centroid , while its dimension is determined to be the 95% error region for 4 d.o.f . for 4 d.o.f
.. there are 4 d.o.f . , because of 4 variables : right ascension , declination , flux normalization and photon index . ] .
we found that the 95% error region of the 1@xmath350 gev centroid covers the entire kes 73/1e 1841 - 045 system , but is mutually exclusive with its two adjacent compact h ii regions , g27.276 + 0.148 and g27.491 + 0.189 .
the best - fit 10@xmath3200 gev centroid is offset from the magnetar / snr pair by @xmath2@xmath22 , but its 95% error region contains the pair .
the spatial coincidence of the 1@xmath350 gev feature lat observed and hess j1841 - 055 is very marginal , but the 1@xmath350 gev centroid determined from lat data still has a detection significance of @xmath2@xmath23 in hess ( cf . * ? ? ?
* ; * ? ? ?
* ) . for examining whether the morphology of the emission from this region is extended
, we followed a scheme adopted by @xcite .
we performed a likelihood ratio test to quantify the significance of extension .
we adopted the position of the 1@xmath350 gev centroid , ( 18:41:07.723 , @xmath304:58:07.44 ( j2000 ) ) , for the 1e 1841 - 045 region ( namely _ fermi _ j1841.1 - 0458 ) .
we assigned it a simple power law , and we attempted uniform - disk morphologies of different radii as well as a point - source model on it .
we took the front " data in 1@xmath3400 gev , which thoroughly covers both he and vhe bands .
the @xmath24@xmath25 of different radii relative to the point - source model are tabulated in table [ ext ] .
the most likely radius is determined to be @xmath26 uncertainties are determined at where the @xmath24@xmath25 is lower than the maximum by 1 according to the chi - square distribution . ] and this morphology is preferred over a point - source model by @xmath2@xmath27 .
we therefore adopted this morphology for _ fermi _ j1841.1 - 0458 in subsequent analyses . in the 3xmm j185246.6 + 003317 region
, we identified another extended source ( namely _ fermi _
j1852.2 + 0037 ) , which was modelled together with _
j1841.1 - 0458 and will be reported in our follow - up paper . within 5@xmath7 from the roi center , we discovered two more sources at ( 281.79408@xmath7 , @xmath32.3781283@xmath7)@xmath28 ( ts@xmath0300 ) and ( 281.41503@xmath7 , @xmath33.0315034@xmath7)@xmath28 ( ts@xmath050 ) . in the same chain of analyses
, we also modelled them as additional point sources , merely for the sake of better subtracting the @xmath5-ray backgrounds . to construct the binned spectrum of _ fermi _ j1841.1 - 0458
, we performed an independent fitting of each spectral bin .
for each spectral bin , we assigned a pl model to _ fermi _ j1841.1 - 0458 . considering that we include photons with energies below 1 gev , and that we are investigating crowded regions in the galactic plane , we find it _ inappropriate _ to adopt a full set of data whose large psf ( e.g. , a 68% containment radius of @xmath23.0@xmath7 at 0.2 gev ; cf .
slac @xmath29 ) leads to severe source confusion .
meanwhile , adopting only psf3 " data is also _ discouraged _ in spectral fittings because of large systematic uncertainties induced by severe energy dispersion . for front " data in 0.2@xmath31 gev , the fwhm of its psf is @xmath175% of that for a full set of data and its energy dispersion is greater than that for a full set of data by only @xmath181% of the photon energy ( cf .
slac @xmath29 ) .
therefore , we adopted only front " data in order to achieve a compromise between a small psf and lessened energy dispersion . the spectral energy distribution ( sed ) is shown in figure [ 1e_sed ] .
noticeably , there appears to be a spectral disconnection at energies around 10 gev , where the flux is lower .
this suggests that the entire lat spectrum may consist of two components , which are decoupled at energies around 10 gev . with regards to this
, we looked into the spectral shapes of two mutually exclusive energy bands respectively : 0.2@xmath310 gev and 10@xmath3200 gev .
we examined how well the spectrum in each band can be described by , respectively , a simple power - law ( pl ) @xmath30 , an exponential cutoff power law ( ple ) @xmath31 , and a broken power law ( bkpl ) @xmath32 .
the results of spectral fitting are tabulated in table [ spectral1 ] . for the 0.2@xmath310 gev spectrum of _ fermi _ j1841.1 - 0458
, the likelihood ratio test indicates that both ple and bkpl are preferred over pl by @xmath0@xmath33 .
a ple model yields a photon index of @xmath34 and a cutoff energy of @xmath35 mev .
a bkpl model yields a photon index @xmath36 below the spectral break @xmath37 mev and a photon index @xmath38 above the break .
since the ts values yielded by ple and bkpl models differ by @xmath16 , these two models are similarly preferable . in the 10@xmath3200 gev spectrum of _ fermi _ j1841.1 - 0458 , a pl model yields a photon index @xmath39 , which is consistent with a flat energy - spectrum within the tolerance of the @xmath40 uncertainty .
the relatively low statistics do not allow distinguishing among the goodness of fit of pl , ple and bkpl models . in order to demonstrate whether this spectrum could be well connected to the 1@xmath310 gev spectrum by one single pl
, we also performed a likelihood ratio test for the 1@xmath3200 gev band .
it turns out that a bkpl model with @xmath41 fixed at 10 gev ( @xmath42 and @xmath43 ) is preferred over the best - fit pl ( @xmath44 ) by @xmath2@xmath45 ( i.e. , the probability that the 1@xmath3200 gev spectrum can be joined by a pl is @xmath20.5% ) . in order to examine the long - term variability of _ fermi _ j1841.1 - 0458 , we divided the first @xmath27.4 years of _ fermi _ lat observation into a number of 180-day segments . a binned maximum - likelihood analysis of front " data again , the reason for adopting only front " data is to achieve a compromise between a small psf and lessened energy dispersion . ] in 0.7@xmath3400 gev
was performed for each individual segment .
the choice of a minimum energy cut of 0.7 gev in this analysis is motivated by the fwhm of the psf at 0.7 gev which is @xmath140% of that at 0.2 gev ( cf .
slac @xmath29 ) . in other words , the source confusion of photons in this analysis
is conspicuously _
less _ serious .
we assumed a pl model for _ fermi _ j1841.1 - 0458 .
we then repeated this procedure with 270 days as a segment .
the temporal behaviors of the photon flux and photon index of _ fermi _ j1841.1 - 0458 are plotted with the x - ray outburst history of 1e 1841 - 045 , taken from @xcite , @xcite and references therein @xmath46 , altogether in figure [ lc ] .
if we adopt 180 days as a bin , constant - value functions are sufficient to describe the temporal distributions of flux and spectral index , with @xmath47 for 14 d.o.f . , and
_ no _ bins deviate from the best - fit flat line by @xmath48@xmath49 . if we change the bin size to be 270 days , the temporal distributions of flux and spectral index are also uniform , with @xmath50 for 9 d.o.f . , and there are _ no _ bins different from the best - fit horizontal line by @xmath48@xmath51 .
more importantly , the errors considered in above fittings are purely statistical . if we take the systematic uncertainties of galactic diffuse emission model and the effective area ( cf . * ? ? ?
* ) into consideration , then the significance of variability will further be lowered .
thus , _ no _ @xmath5-ray variability of _ fermi _ j1841.1 - 0458 is significantly detected .
our analysis of _ fermi _ j1841.1 - 0458 confirms what @xcite and @xcite discovered with _
fermi _ lat data : the presence of a gev - bright extended source which is positionally consistent with the kes 73/1e 1841 - 045 system .
@xcite placed a 95% upper limit of @xmath52 erg @xmath53 s@xmath10 on the 0.1@xmath310 gev flux of the magnetar 1e 1841 - 045 , after their attempt to subtract the @xmath5-ray contribution from snr kes 73 which is enclosing the magnetar at its center .
this upper limit is approximately one - fifth of our observed flux for _ fermi _ j1841.1 - 0458 .
nevertheless , we take a different approach to discuss possible @xmath5-ray origins .
_ fermi _ j1841.1 - 0458 radiates strongly in both low - energy ( le ; 0.1@xmath31 gev ) , he ( 1@xmath310 gev ) and vhe ( @xmath010 gev ) regimes . comprehensively considering figure [ 1e_sed ] and table [ spectral1 ]
, we found that the 0.1@xmath310 gev and 10@xmath3200 gev emissions may be respectively dominated by two components of different mechanisms .
we will therefore discuss the emission mechanisms of the source using these two energy bands .
we compared the lat spectral shape of _ fermi _ j1841.1 - 0458 with those of other snrs younger than or as young as kes 73 , which are tabulated in table [ youngsnr ] .
the interpretations based on this comparison will be discussed in the following sub - sections .
the best - fit 1@xmath350 gev centroid of _ fermi _ j1841.1 - 0458 is almost at the center of kes 73 , while the 95% error region of this centroid is mutually exclusive with the two h ii clouds g27.276 + 0.148 and g27.491 + 0.189 .
this contradicts the mc - interaction scenario where the actual @xmath5-ray source is the proton collision sites ( mcs ) rather than the proton acceleration sites ( snr shocks ) .
whereas , this is within our expectation , as only @xmath25% of cosmic rays accelerated in kes 73 can reach these two h ii clouds , according to the angular sizes of the clouds and their angular separations from kes 73 ( taken from figure 1(a ) of * ? ? ?
also , kes 73 is a shell - type snr which shows _ no _ signs of being thermal composite , further confirming that the dense medium interacting with kes 73 is _ not _ that abundant .
furthermore , a snr is generally gev - brightest at an age within @xmath23@xmath310 kyr ( cf .
* ) , while kes 73 is only @xmath182.1 kyr old @xcite .
therefore , the @xmath5-ray contribution from kes 73 should be limited .
the _ fermi _ lat detection of the gev counterparts of cas a and tycho @xcite , respectively @xmath2340 yr old and @xmath2440 yr old snrs in lack of dense medium in their neighborhoods @xcite , makes it rather possible for kes 73 to be a significant @xmath5-ray source like cas a and tycho . however , if kes 73 ( at a distance of @xmath28.5 kpc ; * ? ? ?
* ; * ? ? ?
* ) is the only source of energy injection for _ fermi _ j1841.1 - 0458 , then its 0.1@xmath31 gev luminosity would be a factor of @xmath020 greater than those of cas a and tycho ( at distances of @xmath23.3 kpc and @xmath15 kpc respectively ; * ? ? ?
? * ; * ? ? ?
besides , the @xmath21@xmath310 gev spectrum of _ fermi _ j1841.1 - 0458 is steeper than those of all other young - snr associated sources in table [ youngsnr ] by @xmath0@xmath54 .
furthermore , @xmath183000 yr old snrs should have @xmath5-ray luminosities @xmath18@xmath55 erg s@xmath10 ( cf . * ? ? ?
* ) , while the 0.1@xmath310 gev luminosity of _ fermi _ j1841.1 - 0458 is @xmath2@xmath56 erg s@xmath10 , based on the bkpl model fit to the 0.2@xmath310 gev spectrum .
if we adopt the simple model where the target material for cosmic - ray interaction at early snr ages is only the shocked interstellar medium but _ not _ the snr shell ( cf .
* ) , then it is _ unusual _ for kes 73 to dominate the le@xmath3he ( 0.1@xmath310 gev ) emission of _ fermi _ j1841.1 - 0458 . in 10@xmath3200 gev , the luminosity of _ fermi _ j1841.1 - 0458 is @xmath2@xmath57 erg s@xmath10 , with a hard photon index of @xmath58 .
interestingly , the 1@xmath3300 gev spectrum of the snr rx j0852.0 - 4622 ( 2400@xmath35100
yr old ; * ? ? ?
* ) is well described by a pl model with a similar photon index of @xmath59 @xcite , and the 0.5@xmath3300 gev spectrum of the snr rx j1713.7 - 3946 ( @xmath21600 yr old ; cf .
* ; * ? ? ?
* ) is well described by a pl model with an even slightly harder photon index of @xmath60 @xcite .
for these two snrs , any simple one - zone model for particle acceleration does not favor a strong hadronic pion - decay component @xcite , as more - energetic particles are more inclined to lose their energy or to escape from the acceleration sites .
such hard gev spectra of snrs can possibly be explained by leptonic cosmic - rays up - scattering soft photons ( cf .
* ; * ? ? ?
* ) , and/or more complicated hadronic emission ( beyond one - zone models ) from a shell of dense gas that is transferred from progenitor stellar winds and is located a short distance upstream of the forward shock ( cf . * ? ? ?
since kes 73 is as young as rx j1713.7 - 3946 and is just slightly younger than rx j0852.0 - 4622 , it is likely that kes 73 dominates the vhe emission of _ fermi _ j1841.1 - 0458 , whose vhe spectrum is as hard as those of rx j0852.0 - 4622 and rx j1713.7 - 3946 . both the gev@xmath3tev spectra of rx j0852.0 - 4622 and rx j1713.7 - 3946 are well described by bkpl models with spectral turnovers and peaks at @xmath2300 gev @xcite .
hess observation in the tev band shows that kes 73 has a detection significance of @xmath2@xmath23 at the northern edge of hess j1841 - 055 ( cf .
* ; * ? ? ?
unfortunately , it is _ inappropriate _ to examine the spectral connection for _ fermi _ j1841.1 - 0458 and hess j1841 - 055 , since hess j1841 - 055 has an extended feature composed of kes 73 as well as several stronger high - energy sources @xcite .
it is unclear whether the vhe spectrum of _ fermi _ j1841.1 - 0458 can be connected to the tev spectrum of kes 73 ( after subtracting other components of hess j1841 - 055 ) with a turnover and peak at @xmath2300 gev , like what is observed in the spectra of rx j0852.0 - 4622 and rx j1713.7 - 3946 .
it is still dubious whether the scenarios explaining the gev spectra of rx j0852.0 - 4622 and rx j1713.7 - 3946 can be applied to kes 73 or not .
1e 1841 - 045 is highly consistent with the 1@xmath350 gev centroid of _ fermi _ j1841.1 - 0458 .
one of the most preferable spectral models in 0.2@xmath310 gev , ple , yields a spectral index within the range of typical values of about 1 - 2 in the 2nd catalog of pulsars detected by the _ fermi _ lat , and a cutoff energy consistent with the range of typical values of about 1 - 4 gev @xcite . while the spin - down power of 1e 1841 - 045 of @xmath2@xmath61 erg s@xmath10 @xcite is only @xmath20.11% of the 0.1@xmath310 gev luminosity of _ fermi _ j1841.1 - 0458 , its surface magnetic field of @xmath2@xmath62 g at the pole ( cf . * ? ? ?
* ) makes it appropriate to estimate the power of magnetic field decay to be @xmath16@xmath63 erg s@xmath10 ( cf .
* ) , which is marginally greater than the 0.1@xmath310 gev luminosity of _ fermi _ j1841.1 - 0458 . therefore , one possible scenario is that the le@xmath3he emission of _ fermi _ j1841.1 - 0458 is dominated by energy released from the very - intense magnetic field of 1e 1841 - 045 .
the @xmath00.7 gev emission of _ fermi _ j1841.1 - 0458 is essentially steady .
both figure [ lc ] in this work and figure 5 in @xcite ( a long - term light - curve of the sgr 1806 - 20 region ) suggest that , regardless of cosmic - ray origins , @xmath5-ray emissions from magnetars regions may _ not _ vary in response to magnetar outbursts , although the enormous energy ejected from these outbursts triggers abrupt yet dramatic enhancements in x - ray and _ soft _ @xmath5-ray emissions . the 10@xmath3200 gev luminosity of _ fermi _ j1841.1 - 0458 is 2 orders of magnitude higher than the spin - down power of 1e 1841 - 045 , but is about an order of magnitude lower than its power of magnetic field decay .
although synchrotron cooling generally makes leptonic cosmic - rays difficult to produce @xmath5-ray photons of a few gev or above via synchrotron radiation , pwne can , with the reduced synchrotron losses for high - energy ic - emitting electrons , maintain their high gev@xmath3tev @xmath5-ray fluxes for timescales exceeding the lifetime of their progenitor pulsars @xcite .
remarkably , in 10@xmath3200 gev , the luminosity and spectral shape of _ fermi _ j1841.1 - 0458 are both highly consistent with those of crab nebula s ic component ( see table [ youngsnr ] , and compare figure [ 1e_sed ] in this work with figure 2 of * ? ? ?
crab nebula is a pwn powered by a @xmath21 kyr old pulsar ( as young as the kes 73/1e 1841 - 045 system ; * ? ? ?
* ) , prompting us to speculate that kes 73 also contains an ic - radiating pwn . on the other hand , a stark contrast between the 0.2@xmath310 gev spectra of _ fermi _ j1841.1 - 0458 and crab nebula ( cf .
table [ youngsnr ] ) makes it _ unreasonable _ to apply the synchrotron component of crab nebula ( for detail , see * ? ? ?
* ) to explain the le@xmath3he emission of _ fermi _ j1841.1 - 0458 .
tev - detected pwne should be associated with pulsars releasing power of @xmath0@xmath63 erg s@xmath10 @xcite .
such required power is greater than the rotational power of 1e 1841 - 045 by three orders of magnitude , but can be afforded by magnetic field decay of 1e 1841 - 045 .
swift j1834.9 - 0846 , the first magnetar confirmed to produce a pwn @xcite , is under a similar circumstance ( cf .
* ) , and hence the wind nebula around it has been argued to be magnetically powered @xcite .
conceivably , magnetar s relativistic outflows , which are accumulated over all outbursts in its history and confined by snr shell , can supply even larger fluxes of wind particles than those supplied in quiescence @xcite .
therefore , 1e 1841 - 045 is sufficient to generate an ic - radiating pwn which may account for a significant portion of the lat flux in the vhe band and/or the hess flux in the tev band .
however , a major uncertainty of this scenario is that there is _ no _ firmly identified pwn inside the snr shell of kes 73 .
subject to assured authenticity of the model prediction for young ( @xmath183 kyr old ) snrs done by @xcite , the le@xmath3he ( 0.1@xmath310 gev ) emission of _ fermi _ j1841.1 - 0458 _ can not _ be dominated by snr kes 73 , and is more likely to be dominated by the magnetar 1e 1841 - 045 . whereas , unless more and more young snrs are found to have low 0.1@xmath310 gev luminosities ( like those of cas a , tycho , rx j0852.0 - 4622 and rx j1713.7 - 3946 ) , the credibility of their model for snr evolution may still be challenged .
once it is challenged , we _ can not _ robustly eliminate the possibility of the kes 73 origin for this band .
the vhe ( @xmath010 gev ) emission of _ fermi _ j1841.1 - 0458 is evidently dominated by kes 73 , while a pwn generated by 1e 1841 - 045 , if it exists , is a sufficient yet unnecessary source in this energy band .
we suggest modelling the kes 73/1e 1841 - 045 system as a single source independent from hess j1841 - 055 in the tev regime , so that we can perform physically meaningful comparisons between gev and tev spectra . in turn
, we can compare its gev@xmath3tev spectrum to those of rx j0852.0 - 4622 , rx j1713.7 - 3946 and crab nebula .
how will the gev and tev analyses of this magnetar / snr pair be affected if we assign a different morphology to hess j1841 - 055 ? what should be the refined position and dimension of hess j1841 - 055 so that it excludes the magnetar / snr pair more precisely while retaining all other high - energy components ? to address such crucial issues , ones can start with analyzing tev data collected by hess , magic and/or cta , each of which has a higher sensitivity than the gev - observing _ fermi _ lat .
this work is supported by the ministry of science and technology of the republic of china ( taiwan ) through grants 103 - 2628-m-007 - 003-my3 and 105 - 2112-m-007 - 033-my2 .
is supported by the one hundred talent program of the sun yat - sen university and the fundamental research funds for the central universities in p. r. china .
is supported by the national research foundation of korea through grants 2014r1a1a2058590 and 2016r1a5a1013277 .
j.t . is supported by nsfc grants of chinese government under 11573010 and u1631103 .
is supported by a grf grant under 17300814 .
thanks x. hou and l. c. c. lin for some useful discussion .
abdo , a. a. , ackermann , m. , ajello , m. , et al .
2009 , apjl , 706 , l1 abdo , a. a. , ackermann , m. , ajello , m. , et al . 2010a , apj , 718 , 348 abdo , a. a. , ackermann , m. , ajello , m. , et al .
2010b , sci , 327 , 1103 abdo , a. a. , ackermann , m. , ajello , m. , et al . 2010c , apj , 712 , 459 abdo , a. a. , ackermann , m. , ajello , m. , et al .
2010d , apjl , 725 , l73 abdo , a. a. , ackermann , m. , ajello , m. , et al . 2010e , apjl , 710 , l92 abdo , a. a. , ajello , m. , allafort , a. , et al .
2013 , apjs , 208 , 17 acero , f. , ackermann , m. , ajello , m. et al .
2015 , apjs , 218 , 23 acero , f. , ackermann , m. , ajello , m. et al .
2016 , apjs , 224 , 8 aharonian , f. , akhperjanian , a. g. , barres de almeida , u. , et al .
2008 , a&a , 477 , 353 alarie , a. , bilodeau , a. , & drissen , l. 2014 , mnras , 441 , 2996 allen , g. e. , chow , k. , delaney , t. , et al .
2015 , apj , 798 , 82 araya , m. 2015 , apj , 813 , 3 archambault , s. , archer , a. , benbow , w. , et al .
2017 , apj , 836 , 23 auchettl , k. , slane , p. , & castro , d. 2014 , apj , 783 , 32 beloborodov , a. m. , & thompson , c. 2007 , apj , 657 , 967 bochow , a. 2011 , phd thesis , ruperto - carola univ .
heidelberg buehler , r. , scargle , j. d. , blandford , r. d. , et al .
2012 , apj , 749 , 26 carter , l. m. , dickel , j. r. , & bomans , d. j. 1997 , pasp , 109 , 990 castro , d. , slane , p. , carlton , a. , & figueroa - feliciano , e. 2013 , apj , 774 , 36 cheng , k. s. , & zhang , l. 2001 , apj , 562 , 918 collazzi , a. c. , kouveliotou , c. , van der horst , a. j. , et al .
2015 , apjs , 218 , 11 dermer , c. d. , & powale , g. 2013 , a&a , 553 , a34 dib , r. , & kaspi , v. m. 2014 , apj , 784 , 37 duncan , r. c. 1998 , apj , 498 , l45 duncan , r. c. , & thompson , c. 1992 , apj , 392 , l9 federici , s. , pohl , m. , telezhinsky , i. , wilhelm , a. , & dwarkadas , v. v. 2015 , a&a , 577 , a12 fesen , r. a. , hammell , m. c. , morse , j. , et al .
2006 , apj , 636 , 859 fesen , r. a. , kremer , r. , patnaude , d. , & milisavljevic , d. 2012 , aj , 143 , 27 gaensler , b. m. , & slane , p. o. 2006 , ara&a , 44 , 17 giordano , f. , naumann - godo , m. , ballet , j. , et al .
2012 , apj , 744 , l2 granot , j. , gill , r. , younes , g. , et al . 2017 , mnras , 464 , 4895 green , a. j. , frail , d. a. , goss , w. m. , & otrupcek , r. 1997 , aj , 114 , 2058 halpern , j. p. , & gotthelf , e. v. 2010 , apj , 725 , 1384 hayato , a. , yamaguchi , h. , tamagawa , t. , et al .
2010 , apj , 725 , 894 helfand , d. j. , zoonematkermani , s. , becker , r. h. , & white , r. l. 1992 , apjs , 80 , 211 kargaltsev , o. , kouveliotou , c. , pavlov , g. g. , et al . , 2012 , apj , 748 , 26 katsuda , s. , acero , f. , tominaga , n. et al .
2015 , apj , 814 , 29 kilpatrick , c. d. , bieging , j. h. , & rieke , g. h. 2014 , apj , 796 , 144 kilpatrick , c. d. , bieging , j. h. , & rieke , g. h. 2016 , apj , 816,1 kumar , h. s. , safi - harb , s. , slane , p. o. , & gotthelf , e. v. 2014 , apj , 781 , 41 li , j. , rea , n. , torres , d. f. , & de ona - wilhelmi , e. 2017 , apj , 835 , 30 lin , l. , kouveliotou , c. , g , e. , et al .
2011 , apjl , 740 , l16 liu , b. , chen , y. , zhang , x. et al .
2015 , apj , 809 , 102 ma , y. k. , ng , c .- y . , bucciantini , n. , et al .
2016 , apj , 820 , 100 olausen , s. a. , & kaspi , v. m. 2014 , apjs , 212 , 6 rudie , g. c. , fesen , r. a. , & yamada , t. 2008 , mnras , 384 , 1200 slysh , v. i. , valtts , i. e. , kalenskii , s. v. , et al . 1999 ,
a&as , 134 , 115 takata , j. , wang , y. , wu , e. m. h. , cheng , k. s. , 2013 , mnras , 431 , 2645 tanaka , t. , allafort , a. , ballet , j. , et al .
2011 , apjl , 740 , l51 tian , w. w. , & leahy , d. a. 2008 , apj , 677 , 292 tian , w. w. , & leahy , d. a. 2011 , apjl , 729 , l15 tibolla , o. , mannheim , k. , elssser , d. , & kaufmann , s. 2011 , arxiv:1111.1634 tong , h. 2016 , raa , 16 , 143 vasisht , g. , & gotthelf , e. v. 1997 , apjl , 486 , l129 wachter , s. , patel , s. k. , kouveliotou , c. , et al .
2004 , apj , 615 , 887 xing , y. , wang , z. , zhang , x. , & chen , y. 2014 , apj , 781 , 64 yeung , p. k. h. , kong , a. k. h. , tam , p. h. t. , et al .
2016 , apj , 827 , 41 younes , g. , kouveliotou , c. , kargaltsev , o. , et al .
2016 , apj , 824 , 138 zhang , b. 2003 , in proc .
workshop on strong magnetic fields and neutron stars : spin - down power of magnetars , ed .
h. j. mosquera cuesta , h. perz rojas , & c. a. zen vasconcellos , 83 zirakashvili , v. n. , aharonian , f. , yang , r. , wilhelmi , e. o. , & tuffs , r. j. 2014 , apj , 785 , 130 50 gev ( for psf3 " data ; left ) and 10@xmath3200 gev ( for a full set of data ; right ) , respectively , where all neighboring 3fgl catalog sources are subtracted .
the green thick circles indicate the position and dimension of kes 73 , which are taken from @xcite .
the black diamonds indicate the position of 1e 1841 - 045 , which is taken from @xcite . the position and dimension of the two h ii clouds g27.276 + 0.148 and g27.491 + 0.189 , indicated as cyan circles ,
are taken from @xcite .
the position and dimension of hess j1841 - 055 , indicated as a gray ellipse , are taken from @xcite .
the positions of nearby 3fgl sources are marked by red crosses .
on the left map , the centroid is indicated as a black thick cross , whose size illustrates the 95% error region.,title="fig : " ] 50 gev ( for psf3 " data ; left ) and 10@xmath3200 gev ( for a full set of data ; right ) , respectively , where all neighboring 3fgl catalog sources are subtracted .
the green thick circles indicate the position and dimension of kes 73 , which are taken from @xcite .
the black diamonds indicate the position of 1e 1841 - 045 , which is taken from @xcite . the position and dimension of the two h ii clouds g27.276 + 0.148 and g27.491 + 0.189 , indicated as cyan circles , are taken from @xcite .
the position and dimension of hess j1841 - 055 , indicated as a gray ellipse , are taken from @xcite .
the positions of nearby 3fgl sources are marked by red crosses .
on the left map , the centroid is indicated as a black thick cross , whose size illustrates the 95% error region.,title="fig : " ] 400 gev light - curve of _ fermi _ j1841.1 - 0458 .
the size of each black solid bin is 180 days , and the size of each gray dashed bin is 270 days .
the red vertical lines indicate the dates of x - ray outbursts of 1e 1841 - 045 , and the shaded time - range contains a series of 10 x - ray outbursts ( for detail , see * ? ? ?
* ; * ? ? ?
* and references therein @xmath46 ) .
( b ) temporal behavior of the 0.7@xmath3400 gev spectral index of _ fermi _ j1841.1 - 0458 with the same binning methods as those applied in ( a ) . ]
.the @xmath24@xmath25 in 1@xmath3400 gev for front " data , when uniform disks of different radii replace the point - source model to be the morphology of _ fermi _ j1841.1 - 0458 . [ cols="<,^",options="header " , ] | the supernova remnant kes 73 and/or the magnetar 1e 1841 - 045 at its center can deposit a large amount of energy to the surroundings and is potentially responsible for particle acceleration . using the data taken with the _ fermi _ large area telescope ( lat )
, we confirmed the presence of an extended source whose centroid position is highly consistent with this magnetar / supernova - remnant pair .
its emission is intense from 100 mev to @xmath0100 gev .
its lat spectrum can be decoupled into two components which are respectively governed by two different mechanisms . according to the young age of this system ,
the magnetar is seemingly a necessary and sufficient source for the downward - curved spectrum below 10 gev , as the observed @xmath110 gev flux is too high for the supernova remnant to account for . on the other hand ,
the supernova remnant is reasonably responsible for the hard spectrum above 10 gev .
further studies of this region in the tev regime is required , so that we can perform physically meaningful comparisons of the @xmath010 gev spectrum and the tev spectrum . |
inverse problems arise whenever one searches for unknown causes based on observation of their effects . driven by the requirements from huge amount of practical applications ,
the field of inverse problems has undergone a tremendous growth .
such problems are usually ill - posed in the sense that their solutions do not depend continuously on the data . in practical applications ,
one never has exact data , instead only noisy data are available due to errors in the measurements . even if the deviation is very small , algorithms developed for well - posed problems may fail , since noise could be amplified by an arbitrarily large factor . therefore , the development of stable methods for solving inverse problems is a central topic . in this paper
we consider the stable resolution of nonlinear inverse problems which mathematically can be formulated as the nonlinear equations @xmath0 where @xmath1 is a nonlinear frchet differentiable operator between two hilbert spaces @xmath2 and @xmath3 whose norms and inner products are denoted as @xmath4 and @xmath5 respectively .
we use @xmath6 to denote the frchet derivative of @xmath7 at @xmath8 and use @xmath9 to denote the adjoint of @xmath6 .
we assume that ( [ 1 ] ) has a solution @xmath10 in the domain @xmath11 of @xmath7 , i.e. @xmath12 . let @xmath13 be the only available noisy data of @xmath14 satisfying @xmath15 with a given small noise level @xmath16 . due to the intrinsic ill - posedness ,
regularization methods should be employed to produce from @xmath13 a stable approximate solution of ( [ 1 ] ) .
many regularization methods have been considered in the last two decades . due to
their straightforward implementation and fast convergence property , newton type regularization methods are attractive for solving nonlinear inverse problems .
in @xcite we considered a general class of newton type methods of the form @xmath17 where @xmath18 is an initial guess of @xmath10 , @xmath19 is a sequence of positive numbers , and @xmath20 is a family of spectral filter functions
. the scheme ( [ m1 ] ) can be derived by applying the linear regularization method defined by @xmath20 to the linearized equation @xmath21 which follows from ( [ 1 ] ) by replacing @xmath14 by @xmath13 and @xmath22 by its linearization @xmath23 at @xmath24 .
when the sequence @xmath19 is given a priori with suitable property , we showed in @xcite that , under the discrepancy principle , the methods are convergent and order optimal .
we also considered in @xcite the methods in hilbert scales and obtained the order optimal convergence rates . in the definition of the newton type methods ( [ m1 ] )
, one may determine the sequence @xmath19 adaptively during computation .
motivated by the inexact newton methods in @xcite for well - posed problems , hanke proposed in @xcite his regularizing levenberg - marquardt scheme for solving nonlinear inverse problems with @xmath19 chosen to satisfy @xmath25 at each step for some preassigned number @xmath26 and with the discrepancy principle used to terminate the iteration .
rieder generalized the idea in @xcite and proposed in @xcite ( see also @xcite ) a general class of inexact newton methods ; every such a method consists of two components : an outer newton iteration and an inner scheme providing increment by regularizing local linearized equations .
when the inner scheme is defined by an iterative method , the number of iterations is determined adaptively which has the advantage to avoid the over - solving of the linearized equation that may occur when the inner scheme is terminated a priori .
the convergence rates of inexact newton regularization methods were considered in @xcite but only suboptimal ones were derived .
it is a longstanding question whether the inexact newton methods are order optimal .
important progress has been made recently in @xcite where the regularizing levenberg - marquardt scheme is shown to be order optimal . in this paper
we consider the inexact newton regularization methods in which the inner schemes are defined by applying various linear regularization methods , including landweber iteration , the implicit iteration , the asymptotic regularization and tikhonov regularization , to the local linearized equations and show that these methods are indeed order optimal by exploiting ideas developed in @xcite .
we even consider these methods in hilbert scales and derive the order optimal convergence rates .
our theoretical results confirm those numerical illustrations in @xcite .
this paper is organized as follows . in section 2
we formulate the methods precisely and state the main results on the order optimal convergence rates . in section 3
we show that these methods are well - defined , and prove that the error decays monotonically . in section 4
we complete the proof of the the main result by deriving the order optimal convergence rates .
the inexact newton regularization methods are a family of methods for solving nonlinear ill - posed inverse problems .
every such a method consists of two components , an outer newton iteration and an inner scheme providing increments by regularizing local linearized equations .
an approximate solution is output by a discrepancy principle . to be more precise ,
the method starts with an initial guess @xmath27 .
assume that @xmath24 is a current iterate , one may apply any regularization scheme to the linearized equation @xmath28 to produce a family of regularized approximations @xmath29 .
one may choose @xmath30 to be the smallest number @xmath31 such that @xmath32 for some preassigned value @xmath33 .
the next iterate is then updated as @xmath34 .
the outer newton iteration is terminated by the discrepancy principle @xmath35 for some given number @xmath36 .
this outputs an integer @xmath37 and hence @xmath38 which is used to approximate the exact solution @xmath10 .
the convergence rates of the inexact newton regularization methods have been considered in @xcite .
it has been shown that if @xmath39 for some @xmath40 , then there is a number @xmath41 such that @xmath42 which is only suboptimal .
it is a long - standing question whether the inexact newton regularization methods are order optimal .
important progress has been made recently in @xcite where the regularizing levenberg - marquardt scheme is proved to be order optimal . in this paper
we will consider the inexact newton regularization methods in which the inner schemes are defined by applying landweber iteration , the implicit iteration , the asymptotic regularization , or tikhonov regularization to the linearized equation ( [ linear ] ) and show that these methods are indeed order optimal .
for these four methods , @xmath43 are defined by @xmath44 with the spectral filter functions @xmath20 given by @xmath45 - 1 } ( 1-\lambda)^j , \,\,\,\ , \sum_{j=1}^{[t ] } ( 1+\lambda)^{-j } , \,\,\,\ ,
\frac{1}{\la } \left(1-e^{-t\la}\right ) , \,\,\,\ , \left(\frac{1}{t}+\la\right)^{-1}\ ] ] respectively , where @xmath46 $ ] denotes the largest integer not greater than @xmath47 .
we need the following standard condition which is known as the newton - mysovskii condition ( see @xcite ) .
[ a0 ] ( a ) there exists @xmath48 such that @xmath49 h\|\le k_0\|x - z\| \|f'(z ) h\| , \quad \forall h\in \x\ ] ] for all @xmath50 , where @xmath51 denotes the ball of radius @xmath52 with center at @xmath10 .
\(b ) @xmath7 is properly scaled so that @xmath53 for all @xmath54 .
the order optimality of these four inexact newton regularization methods is contained in the following result .
[ t1 ] let @xmath7 satisfy assumption [ a0 ] , let @xmath55 and @xmath33 be such that @xmath56 , and let @xmath57 . if @xmath58 is sufficiently small , then the inexact newton regularization methods with the inner scheme defined by landweber iteration , the implicit iteration , the asymptotic regularization , or tikhonov regularization are well - defined and terminate after @xmath59 iterations . if , in addition , @xmath60 for some @xmath61 and @xmath40 and if @xmath62 is sufficiently small , then there holds @xmath63 for some constant @xmath64 independent of @xmath65 and @xmath66
. we will not give the proof of theorem [ t1 ] directly .
instead , we will prove a more general result by considering these four inexact newton regularization methods in hilbert scales .
let @xmath67 be a densely defined self - adjoint strictly positive linear operator in @xmath2 satisfying @xmath68 for some constant @xmath69 , where @xmath70 denotes the domain of @xmath67 .
for each @xmath71 , we define @xmath72 to be the completion of @xmath73 with respect to the hilbert space norm @xmath74 this family of hilbert spaces @xmath75 is called the hilbert scales generated by @xmath67 .
the following are fundamental properties ( see @xcite ) : \(a ) for any @xmath76 , @xmath77 is densely and continuously embedded into @xmath78 with @xmath79 \(b ) for any @xmath80 there holds the interpolation inequality @xmath81 \(c ) if @xmath82 is a bounded linear operator satisfying @xmath83 for some constants @xmath84 and @xmath85 , then for the operator @xmath86 with @xmath87 there holds for any @xmath88 that @xmath89 on @xmath90 , where @xmath91 is the adjoint of @xmath92 and @xmath93 we will consider the inexact newton regularization methods in which the inner schemes are defined by applying landweber iteration , the implicit iteration , the asymptotic regularization , or tikhonov regularization in hilbert scales to the linearized equation ( [ linear ] ) .
now we have @xmath94 with @xmath95 defined by ( [ 5.7.2 ] ) , where @xmath96 is a suitable chosen number .
the iterative solutions are defined by @xmath97 with @xmath31 chosen to be the smallest number satisfying ( [ 4.2 ] ) .
the iteration is then terminated by the discrepancy principle ( [ 4.3 ] ) to output an approximate solution @xmath38 .
we will use @xmath38 , constructed from these four inexact newton regularization methods in hilbert scales , to approximate the true solution @xmath10 of ( [ 1 ] ) and derive the order optimal convergence rate when @xmath98 with @xmath99 .
we need the following condition on the nonlinear operator @xmath7 .
[ a1 ] ( a ) there exist constants @xmath85 and @xmath100 such that @xmath101 for all @xmath54 .
\(b ) @xmath7 is properly scaled so that @xmath102 for all @xmath54 , where @xmath87 .
\(c ) there exist @xmath103 , @xmath104 and @xmath48 such that @xmath105 for all @xmath106 .
this condition was first used in @xcite for the convergence analysis of the nonlinear landweber iteration in hilbert scales .
it was then used recently in @xcite and @xcite for nonlinear tikhonov regularization and some newton - type regularization methods in hilbert scales respectively .
one can consult @xcite for several examples satisfying assumption [ a1 ] .
[ t2 ] let @xmath7 satisfy assumption [ a1 ] with @xmath107 , let @xmath55 and @xmath33 be such that @xmath56 , and let @xmath27 be such that @xmath108 .
if @xmath109 is sufficiently small , then the inexact newton regularization methods with the inner scheme defined by landweber iteration , the implicit iteration , the asymptotic regularization , or tikhonov regularization in hilbert scales are well - defined and terminate after @xmath59 iterations . if , in addition , @xmath110 for some @xmath111 and @xmath112 is sufficiently small , then there holds @xmath113 for all @xmath114 $ ] , where @xmath64 is a constant independent of @xmath65 and @xmath115 .
the proof of theorem [ t2 ] will be given in the next two sections . here
some remarks are in order .
when the inner scheme is defined by the asymptotic regularization or tikhonov regularization , there is flexibility to choose @xmath30 to satisfy @xmath116 with some numbers @xmath117 .
furthermore , we only need @xmath55 and @xmath118 in the convergence analysis . when @xmath119 , the same order optimal convergence rate in theorem [ t2 ] holds for @xmath98 with @xmath120 which can be seen from the proof of lemma [ p1 ] in section 4 .
if the frchet derivative @xmath6 satisfies the lipschitz condition @xmath121 then assumption [ a1 ] ( c ) holds with @xmath122 and @xmath123 , and thus , for these inexact newton regularization methods in hilbert scales with @xmath124 , the order optimal convergence rates hold for @xmath98 with @xmath125 .
we indicate how theorem [ t1 ] can be derived from theorem [ t2 ] .
first , we note that assumption [ a0 ] ( a ) implies @xmath126 for all @xmath106 .
one can then follow the proofs in section 3 to show that , if @xmath57 and @xmath58 is sufficiently small , then these inexact newton regularization methods are well - defined and @xmath127 which implies @xmath128 for @xmath129 . by shrinking the ball @xmath51 if necessary , we can derive from assumption [ a0 ] ( a ) that there exist two constants @xmath130 such that @xmath131 for all @xmath106 .
this implies that all the operators @xmath6 have the same null space @xmath132 as long as @xmath54 . by the condition of theorem [ t1 ] we have @xmath133 . by the definition of @xmath134 we also have @xmath135 for @xmath136 . by considering the operator @xmath137 if necessary
, we may assume @xmath138 .
therefore @xmath139 for @xmath140 , and we may consider the equation ( [ 1 ] ) on @xmath141 .
consequently we may assume @xmath142 , i.e. each @xmath6 is injective for @xmath54 .
now we introduce the operator @xmath143 which is clearly densely defined self - adjoint strictly positive linear operator in @xmath2 satisfying @xmath144 from ( [ 5.2.1 ] ) it follows that @xmath145 which implies assumption [ a1 ] ( a ) with @xmath146 . moreover , from assumption [ a0 ] ( b ) it follows for @xmath147that @xmath148 \|_{\x_{-1}\to \y}=\|[f'(x)-f'(z ) ] l\|_{\x\to \y}\le k_0\|x - z\|\|f'(z ) l\|_{\x\to \y}.\end{aligned}\ ] ] since ( [ 5.2.1 ] ) implies @xmath149 , assumption [ a1 ] ( c ) holds with @xmath150 and @xmath123 . since @xmath151 , theorem [ t1 ] follows immediately from theorem [ t2 ] with @xmath152 .
we start with a simple consequence of assumption [ a1 ] which will be used frequently .
[ l2.0 ] let @xmath7 satisfy assumption [ a1 ] and let @xmath106 .
if @xmath153 then @xmath154 if , in addition , @xmath155 , then @xmath156 from assumption [ a1 ] ( c ) and the identity @xmath157 ( x - z ) dt\ ] ] it follows immediately that @xmath158 with the help of the interpolation inequality ( [ inter ] ) we have @xmath159 this together with ( [ 5.1.3 ] ) gives ( [ 5.1.1 ] ) . if , in addition , @xmath155 , then we have @xmath160/(a+t)\ge 1 $ ] . thus , by using @xmath161 which follows from the embedding ( [ embed ] ) , we can derive ( [ 5.1.2 ] ) immediately from ( [ 5.1.1 ] ) .
@xmath162 in this section we will use the ideas from @xcite to show that the four inexact newton regularization methods in hilbert scales stated in theorem [ t2 ] are well - defined and for the error term @xmath163 there holds @xmath164 for @xmath136 .
we will use the notation @xmath165 it follows easily from the definition ( [ 5.7.1 ] ) of @xmath29 that @xmath166 and @xmath167 where @xmath168 denotes the residual function associated with @xmath95 . for the spectral filter functions given in ( [ 5.7.2 ] ) , it is easy to see that @xmath169 for each @xmath170 .
this implies that @xmath171 where @xmath172 denotes the orthogonal projection of @xmath3 onto @xmath173 , the orthogonal complement of the range @xmath174 of @xmath175 .
[ l2.1 ] let @xmath7 satsify assumption [ a1 ] with @xmath107 , let @xmath36 and @xmath33 satisfy @xmath176 , and let @xmath27 be such that @xmath177 .
assume that @xmath178 is sufficiently small . if @xmath179 and @xmath180 , then @xmath30 is well - defined and @xmath181 for some constant @xmath182 independent of @xmath183 and @xmath65 . from ( [ embed ] ) and the given conditions it follows that @xmath184 which implies @xmath128 .
since @xmath185 implies @xmath186 , we have @xmath187 in order to show that @xmath30 is well - defined , in view of ( [ 5.7.4 ] ) it suffices to show @xmath188 since @xmath107 , we can use ( [ 1.2 ] ) and ( [ 5.1.2 ] ) in lemma [ l2.0 ] to derive @xmath189 now by using assumption [ a1 ] ( a ) , @xmath180 and @xmath190 , we obtain with @xmath191 $ ] that @xmath192 since @xmath193 , we therefore obtain ( [ 2.10 ] ) if @xmath194 is sufficiently small . for the inner scheme defined by landweber iteration or the implicit iteration in hilbert scales , it is obvious that @xmath30 is an integer with @xmath195 . for the inner scheme defined by the asymptotic regularization or tikhonov regularization in hilbert scales , we have @xmath196 where @xmath197 or @xmath198 . since @xmath199 , we can obtain either @xmath200 or @xmath201 .
therefore @xmath202 or @xmath203 . @xmath162 [ l2.2 ] let @xmath7 satisfy assumption [ a1 ] with @xmath107 , let @xmath55 and @xmath33 be such that @xmath56 , and let @xmath27 be such that @xmath204 . if @xmath178 is sufficiently small , then the four inexact newton regularization methods in hilbert scales stated in theorem [ t2 ] are well - defined and terminate after @xmath205 iterations , and @xmath206 for some constant @xmath207 .
moreover @xmath208 for @xmath136 .
we will prove this result for the four inexact newton methods case by case .
\(a ) we first consider the inexact newton method with inner scheme defined by landweber iteration in hilber scales .
we first show the monotonicity ( [ 2.11 ] ) .
we may assume @xmath209 .
let @xmath210 and assume that @xmath180 . by the definition of @xmath37 we have @xmath179 .
it follows from lemma [ l2.1 ] that @xmath30 is a well - defined positive integer .
let @xmath211 for each integer @xmath212 . then @xmath213 and @xmath214 for @xmath215 .
recall that @xmath216 .
therefore , in order to show @xmath164 , it suffices to show @xmath217 we set @xmath218 .
then @xmath219 and thus @xmath220 according to the definition of @xmath221 one can see @xmath222 therefore @xmath223 observing that ( [ r1 ] ) and @xmath224}$ ] imply @xmath225 , we have @xmath226 .
hence @xmath227 since @xmath228 , we can pick @xmath229 with @xmath230 . by using assumption [ a1 ] , @xmath190 and @xmath180
, we can derive as in the proof of lemma [ l2.1 ] that if @xmath178 is sufficiently small then @xmath231 on the other hand , by the definition of @xmath30 we have @xmath232
. therefore @xmath233 where @xmath234 .
this in particular implies ( [ 2.12 ] ) and hence @xmath164 . an induction argument then shows the monotonicity result ( [ 2.11 ] ) .
moreover , it follows from ( [ 2.13 ] ) that @xmath235 consequently @xmath236 which shows ( [ 4.30.1 ] ) . since @xmath195 and @xmath237 for @xmath210
, one can see that @xmath37 must be finite .
\(b ) for the inexact newton method with inner scheme defined by the implicit iteration in hilbert scales , all @xmath30 must be positive integer and with the notation @xmath211 we have @xmath213 and @xmath238 let @xmath239 .
we have from ( [ r1 ] ) and @xmath240}$ ] that @xmath241 and @xmath242 .
thus @xmath243 note that @xmath244 .
we then obtain @xmath245 by using @xmath199 and the definition of @xmath30 , we have @xmath246 since @xmath228 , we can obtain @xmath247 for @xmath215 when @xmath194 is sufficiently small , where @xmath229 is such that @xmath230 .
this together with an induction argument implies ( [ 4.30.1 ] ) and ( [ 2.11 ] ) .
\(c ) for the inexact newton method with inner scheme defined by the asymptotic regularization in hilbert scales , @xmath43 is the solution of the initial value problem @xmath248 therefore , with @xmath249 we have @xmath250 according to the definition of @xmath30 we have @xmath251 and @xmath252 for @xmath253 . since @xmath193 , we therefore obtain @xmath254 if @xmath178 is sufficiently small , where @xmath255 is such that @xmath230 . in view of @xmath256 and @xmath257
, we obtain @xmath258 this implies ( [ 4.30.1 ] ) and ( [ 2.11 ] ) immediately .
\(d ) for the inexact newton method with inner scheme defined by tikhonov regularization , we have @xmath259 we first observe that @xmath260 let @xmath261 .
we have from ( [ r1 ] ) and @xmath262 that @xmath263 and hence @xmath264
. therefore @xmath265 by the definition of @xmath30 we have @xmath266 . since @xmath193
, we can obtain @xmath267 if @xmath178 is sufficiently small , where @xmath255 is such that @xmath230 .
this implies ( [ 4.30.1 ] ) and ( [ 2.11 ] ) .
@xmath162 [ r3.1 ] the inequality ( [ 4.30.1 ] ) will find its use in the proof of lemma [ p1 ] . from ( [ 4.30.1 ] ) , @xmath268 , and the fact @xmath269 for @xmath210 , it follows easily that @xmath270 which gives only a rough estimate on the number of outer iterations .
however , we should point out that the inexact newton iterations in hilbert scales in fact terminate after @xmath59 outer iterations .
this can be confirmed by using the fact @xmath271 which follows from the definition of @xmath30 and @xmath257 . to see this , by using ( [ 5.1.2 ] ) in lemma [ l2.0 ] we have @xmath272 since ( [ 2.11 ] ) implies @xmath273 , from assumption [ a1 ] ( a ) we have with @xmath274 $ ] that @xmath275 therefore , if @xmath178 is sufficiently small , then there holds @xmath276 and consequently @xmath277 combining this with ( [ 4.29.0 ] ) yields @xmath278 considering @xmath279 , this in particular implies that if @xmath178 is sufficiently small then @xmath280 therefore for all @xmath140 there holds @xmath281 by taking @xmath282 and using @xmath283 we obtain @xmath284 which shows that @xmath285 .
in this section we will show the order optimality of the four inexact newton method in hilbert scales stated in theorem [ t2 ] .
for simplicity of further exposition , we will always use @xmath64 to denote a generic constant independent of @xmath65 and @xmath183 , we will also use the convention @xmath286 to mean that @xmath287 for some generic constant @xmath64 when the explicit expression of @xmath64 is not important . furthermore , we will use @xmath288 to mean that @xmath286 and @xmath289 .
[ l2.3 ] under the same conditions in lemma [ l2.2 ] , there holds @xmath290 we first claim that there is a constant @xmath291 such that @xmath292 this is clear from the definition of @xmath30 when the inner scheme is defined by tikhonov regularization or the asymptotic regularization .
when the inner scheme is defined by landweber iteration , we have @xmath224}$ ] . according to the definition of @xmath30 and ( [ r1 ] ) , we have @xmath293 since @xmath294
, we have @xmath295 .
therefore , using ( [ r1 ] ) again it follows @xmath296 which shows ( [ 5.7.6 ] ) with @xmath297 .
when the inner scheme is defined by the implicit iteration , we have @xmath240}$ ] .
thus it follows from ( [ r1 ] ) and @xmath199 that @xmath298 which shows ( [ 5.7.6 ] ) with @xmath299 .
the combination of ( [ 5.7.6 ] ) and ( [ 4.29 ] ) gives @xmath300 this shows the result if @xmath178 is sufficiently small .
@xmath162 for the spectral filter functions defined by ( [ 5.7.2 ] ) , we have shown in @xcite that for any sequence of positive numbers @xmath19 there hold @xmath301 and @xmath302 for @xmath303 , @xmath304 and @xmath305 , where @xmath306 is defined by @xmath307 moreover , we have the following crucial estimate .
[ l10 ] let @xmath7 satisfy assumption [ a1 ] , let @xmath20 be defined by ( [ 5.7.2 ] ) and @xmath308 , and let @xmath19 be a sequence of positive numbers with @xmath306 defined by ( [ s1 ] ) .
let @xmath309 and for any @xmath54 let @xmath310 .
then for @xmath311 there holds @xmath312 \right\| \nonumber\\ & \qquad\qquad \qquad\qquad\qquad\qquad \lesssim t_j ( s_n - s_j)^{-\nu-\frac{b+s}{2(a+s ) } } k_0\|x - x^\dag\|^\beta\end{aligned}\ ] ] for @xmath305 .
we refer to ( * ? ? ?
* lemma 2 ) in which similar estimates have been derived for a general class of spectral filter functions .
we also need the following estimate concerning the sums of suitable types which will occur in the convergence analysis .
[ l2 ] let @xmath19 be a sequence of numbers satisfying @xmath313 , and let @xmath314 be defined by ( [ s1 ] ) .
let @xmath315 and @xmath316 be two numbers .
then @xmath317 where @xmath318 is a constant depending only on @xmath319 , @xmath320 and @xmath321 .
this is essentially contained in ( * ? ? ?
* lemma 4.3 ) and its proof .
a simplified proof can be found in ( * ? ? ?
* lemma 3 ) .
@xmath162 now we are ready to give the crucial estimates on @xmath322 and @xmath323 for @xmath210 .
we will exploit the ideas developed in @xcite .
[ p1 ] let @xmath7 satisfy assumption [ a1 ] with @xmath107 , let @xmath55 and @xmath324 be such that @xmath228 , let @xmath27 satisfy @xmath177 . if @xmath325 for some @xmath111 and if @xmath326 is sufficiently small , then there exists a constant @xmath327 such that @xmath328 for all @xmath136 .
since @xmath111 , from ( [ 2.3 ] ) we have @xmath329 .
therefore , it suffices to show that there exists a constant @xmath327 such that @xmath330 for all @xmath136
. we will show ( [ 402 ] ) by induction . by using ( [ 2.3 ] ) and
assumption [ a1 ] ( b ) we have @xmath331 and @xmath332 therefore ( [ 402 ] ) with @xmath333 holds for @xmath334 .
now we assume that ( [ 402 ] ) is true for all @xmath335 for some @xmath336 and want to show that it is also true for @xmath337 . from the equation ( [ g1.1 ] ) and
@xmath257 it follows that @xmath338 ( y^\d - f(x_n)).\end{aligned}\ ] ] by induction on this equation we obtain @xmath339 \left(y^\d - f(x_j)\right).\end{aligned}\ ] ] by multiplying ( [ 20 ] ) by @xmath340 , noting that @xmath341 , and using the identity @xmath342 which follows from the relation @xmath308 , we can obtain @xmath343(y^\d - y ) \nonumber\\ & \quad\ , + \sum_{j=0}^{l-1 } \prod_{k = j+1}^{l-1 } r_{t_k}(aa^ * ) g_{t_j}(aa^ * ) aa^*\left(y - f(x_j)+ t e_j\right ) \nonumber\\ & \quad\ , + \sum_{j=0}^{l-1 } a \prod_{k = j+1}^{l-1 } r_{t_k}(a^*a ) \left[g_{t_j}(a_j^*a_j)a_j^*-g_{t_j}(a^*a)a^*\right ] ( y^\d - f(x_j)).\end{aligned}\ ] ] since @xmath344 with @xmath99 , by using ( [ 2.3 ] ) , ( [ g1 ] ) , ( [ g2 ] ) , ( [ g3 ] ) and lemma [ l10 ] we can derive from ( [ 20 ] ) that @xmath345 where @xmath346 and @xmath64 is a generic constant independent of @xmath347 and @xmath65 . next by using again @xmath344 with @xmath99 , ( [ 2.3 ] ) and ( [ g1 ] ) , we can obtain @xmath348 therefore , it follows from ( [ 21 ] ) , ( [ g2 ] ) and lemma [ l10 ] that @xmath349 we first use ( [ 333 ] ) to derive the desired estimate for @xmath350 .
according to the relation @xmath351 , we have from the induction hypotheses that @xmath352 we need to estimate the terms @xmath353 for each term we will give two types of estimates , one is true for all @xmath354 and the other is true for @xmath355 . by using ( [ 5.1.2 ] ) in lemma
[ l2.0 ] , assumption [ a1 ] ( a ) , lemma [ l2.2 ] , and @xmath356 for @xmath357 we have @xmath358 this shows for @xmath357 that @xmath359 the inequalities ( [ 151 ] ) , ( [ 152 ] ) and lemma [ l2.3 ] imply that if @xmath178 is sufficiently small then @xmath360 consequently , we have from ( [ 151 ] ) and ( [ 212 ] ) that @xmath361 this together with ( [ induction ] ) gives @xmath362 next we estimate @xmath363 .
we have from ( [ 5.1.2 ] ) in lemma [ l2.0 ] , assumption [ a1 ] ( a ) , and ( [ induction ] ) that @xmath364 therefore , it follows from ( [ 212 ] ) that @xmath365 on the other hand , by using ( [ 5.1.1 ] ) in lemma [ l2.0 ] and assumption [ a1 ] ( a ) , we have @xmath366 therefore , it follows from ( [ 212 ] ) and ( [ induction ] ) that @xmath367 for the term @xmath368 , we first have from the interpolation inequality ( [ inter ] ) , lemma [ l2.2 ] , and assumption [ a1 ] ( a ) that @xmath369 with the help of ( [ 152 ] ) we then obtain @xmath370 on the other hand , by using the interpolation inequality ( [ inter ] ) and assumption [ a1 ] ( a ) we also obtain for @xmath354 that @xmath371 this together with ( [ 212 ] ) and ( [ induction ] ) gives @xmath372 now we use ( [ 3.18.1 ] ) , ( [ 214 ] ) and ( [ 216 ] ) with @xmath373 and use ( [ 213 ] ) , ( [ 215 ] ) and ( [ 217 ] ) for @xmath355 , we then obtain from ( [ 333 ] ) that @xmath374 since @xmath375 , we can use lemma [ l2 ] to derive that @xmath376 recall that ( [ 4.30.1 ] ) in lemma [ l2.2 ] implies @xmath377 . since @xmath107 and @xmath378
, we have @xmath379 therefore , noting @xmath380 , we obtain @xmath381 since @xmath382 , we have from the definition of @xmath37 and ( [ 151 ] ) that @xmath383 combining this with ( [ 218 ] ) gives @xmath384 recall that @xmath55 . therefore ,
if @xmath326 is sufficiently small , then we have @xmath385 since @xmath386 and @xmath387 , we have @xmath388 .
therefore @xmath389 if we choose @xmath390 . finally we will use ( [ e29 ] ) to show the desired estimate for @xmath391 .
since we have verified the estimates for @xmath392 , the estimates ( [ 213 ] ) , ( [ 215 ] ) and ( [ 217 ] ) therefore can be improved to include @xmath373 ; this is clear from the above argument .
consequently we have from ( [ e29 ] ) that @xmath393 it then follows from lemma [ l2 ] that @xmath394 with the help of ( [ 219 ] ) and the estimate on @xmath350 , we obtain @xmath395 since @xmath55 , we thus obtain @xmath396 for any @xmath397 if @xmath326 is sufficiently small .
the proof is therefore complete .
@xmath162 now we are ready to complete the proof of theorem [ t2 ] , the main result in this paper .
0.3 cm _ proof of theorem [ t2 ] .
_ considering lemma [ l2.2 ] and remark [ r3.1 ] , it remains only to derive the order optimal convergence rates .
when @xmath398 , the proof is standard .
so we may assume @xmath399 . from lemma [ p1 ]
it follows that @xmath400 . by using lemma [ l2.3 ] and the definition of @xmath37 we have @xmath401 , which together with ( [ 152 ] ) implies that @xmath402
therefore , from the interpolation inequality ( [ inter ] ) it follows that @xmath403 in view of ( [ 2.11 ] ) in lemma [ l2.2 ] , we consequently obtain @xmath404 . by using the definition of @xmath37 and ( [ 1.2 ] ) we have @xmath405 .
observing that ( [ 5.1.2 ] ) in lemma [ l2.0 ] and ( [ 2.11 ] ) in lemma [ l2.2 ] imply @xmath406 thus , if @xmath407 is sufficiently small , then @xmath408 .
consequently @xmath409 .
now we can use again the interpolation inequality ( [ inter ] ) to derive for all @xmath410 $ ] that @xmath411 the proof is therefore complete .
inexact newton regularization methods have been suggested by hanke and rieder in @xcite and @xcite , respectively , for solving nonlinear ill - posed inverse problems .
the convergence rates of these methods have been considered in @xcite , the results however turned out to be inferior to the so - called order optimal rates . for a long time
it has been an open problem whether these inexact newton methods are order optimal , although the numerical illustrations in @xcite present strong indication .
important progress has been made recently in @xcite where the regularizing levenberg - marquardt scheme is shown to be order optimal affirmatively . in this paper
we considered a general class of inexact newton methods in which the inner schemes are defined by landweber iteration , the implicit iteration , the asymptotic regularization and tikhonov regularization . by establishing the monotonicity of iteration errors and deriving a series of subtle estimates , we succeeded in proving the order optimality of these methods .
we also extended these order optimality results to a more general situation where the inner schemes are defined by linear regularization methods in hilbert scales .
our theoretical findings confirm the numerical results in @xcite .
+ * acknowledgement . *
part of the work was carried out during the stay in department of mathematics at virginia tech .
p. deuflhard , h.w .
engl and o. scherzer , _ a convergence analysis of iterative methods for the solution of nonlinear ill - posed problems under affinely invariant conditions _ ,
inverse problems , 14 ( 1998 ) , 10811106 . | inexact newton regularization methods have been proposed by hanke and rieder for solving nonlinear ill - posed inverse problems .
every such a method consists of two components : an outer newton iteration and an inner scheme providing increments by regularizing local linearized equations . the method is terminated by a discrepancy principle . in this paper we consider the inexact newton regularization methods with the inner scheme defined by landweber iteration , the implicit iteration , the asymptotic regularization and tikhonov regularization . under certain conditions
we obtain the order optimal convergence rate result which improves the suboptimal one of rieder .
we in fact obtain a more general order optimality result by considering these inexact newton methods in hilbert scales .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore [ section ] # 1#1_2 # 1#1 @=12 |
in heavy - ion ( ) as well as in proton - proton ( ) collisions , the measurements related to strange hadrons constitute unique tools to study the physics of the strong interaction . in this respect ,
given their strangeness content , the charged multi - strange baryons
( @xmath3 ) , ( @xmath4 ) , ( @xmath5 ) , ( @xmath6 ) are of certain importance . the interest in strangeness and specifically in and can be explained mainly by two reasons : * the initial system formed by the projectiles is free from strange valence quarks , thus the strange quarks that compose the strange hadrons of the final state have to be produced during the process of the collisions ; * due to the identification via weak - decay topology reconstruction , the multi - strange baryons can be studied over a large momentum range , typically from @xmath7 @xmath8 up to @xmath9 .
the resulting @xmath7 spectra then cover the region dominated by the soft processes and reach the energy scale where hard scattering mechanisms may prevail .
a significant part of the investigation in physics depends on our understanding of the production mechanisms ( soft and hard ) in the system , meaning as they occur in the benchmark system .
the measurement of multi - strange particle spectra in collisions is the topic of this report .
the data presented here are from a minimum bias ( mb ) sample of collisions at @xmath0 = 7 , collected at the lhc @xcite with the alice experiment @xcite during summer 2010 .
the entire statistics presently analysed stands for about @xmath10 mb interactions .
the study makes use of the two alice main tracking detectors placed at mid - rapidity , covering the full azimuth : the inner tracking system , composed of 6 cylindrical layers of high - resolution silicon detectors @xcite , and the cylindrical time projection chamber ( tpc ) @xcite .
the multi - strange hadron identification is performed using a combination of displaced - vertex finding , invariant mass analysis and particle identification ( pid ) method at the single track level ( in this analysis , compatibility selection based on the energy loss measurements from the tpc ) .
the reconstruction of the , , , particles hinges on their respective charged weak decays , the so - called _ cascade _ structures . for each species of interest ,
the main characteristics and utilized decay channels are listed in table [ tab : maire - cascdecaychannel ] .
the anti - baryons are similarly reconstructed via the decay channel involving the charge conjugates .
lccccc particles & mass ( @xmath11 ) & @xmath12 ( ) & charged decay channel & b.r .
+ @xmath13 ( @xmath14 ) & @xmath15 & @xmath16 & @xmath17 & 63.9% + @xmath18 ( @xmath3 ) & @xmath19 & @xmath20 & @xmath21 & 99.9% + @xmath22 ( @xmath5 ) & @xmath23 & @xmath24 & @xmath25 & 67.8% + the guidelines of the reconstruction algorithm , consisting of pairing a or baryon with an additional particle ( _ v0 _ structure combined with a so - called _ bachelor _ track ) , are sketched in @xcite and detailed in @xcite .
the protocol to extract the signal counts per @xmath7 interval ( bin ) is also presented in @xcite .
due to the large statistics available for the 7 data sample , and but also and can be studied distinctively . at central rapidity ( ) , the overall raw signal amounts to about @xmath26 counts for , as well as for , and about @xmath27 counts for as well as for its anti - particle .
the upper part of the figure [ fig : maire - fig1-correctedspectra ] shows the four corrected spectra .
they are normalised to the number of inelastic events ( inel ) .
both spectra range from @xmath7 = 0.6 to 8.5 , while the spectra are measured between 0.8 and 5.0 . in order to extract the integrated yields ,
the data points are fitted with a tsallis function @xcite .
the fits result in a good description of data ( @xmath28 close to unity ) and are further used to extrapolate the spectra in the non - measured low @xmath7 region ( @xmath29 of the total @xmath30 for or , @xmath31 for or ) . as a function of transverse momentum , for , , , measured at central rapidity ( @xmath32 ) in inelastic events at @xmath0 = 7 .
the upper panel shows the corrected spectra for the four different species together with their fits by a tsallis function .
the lower panel presents the ratio between the measured data and the predictions returned by the tune perugia 2011 ( p2011 ) of pythia . ] note that , provided the difference in the normalisation scheme chosen by alice ( inel ) and cms ( non - single diffractive , nsd ) @xcite , the ( ) spectra by both experiments are found to be in agreement within the limit of uncertainties .
a significant part of our understanding of collisions is based on description by the monte carlo models ( mc ) .
this is especially the case in the _ soft _ regime ( @xmath7@xmath33 2 - 3 ) .
the confrontation with experimental measurements is necessary , as it spurs further improvements of such phenomenological approaches .
the lower part of the figure [ fig : maire - fig1-correctedspectra ] shows the ratio between data and mc predictions .
the data are here compared to the tune perugia 2011 ( tune s350 @xcite ) of the pythia model @xcite .
although this specific tune provides an improved description of data with respect to earlier tunes ( tunes z2 and perugia 0 ) @xcite , it can still be seen that mc underestimates the measured spectra , up to a factor @xmath34 2 for the charged , @xmath34 5 for the
. however it should be noted for ( ) that there is an agreement between mc and data at @xmath7 @xmath35 6 , where the fragmentation regime may already be reached .
the figure [ fig : maire - fig2-ratioomegaoverxi - pt ] provides another viewpoint on the comparison of data to mc .
it computes the ratio of ( ) to ( ) as a function of @xmath7 . as before ,
it can be noted the model tends to underpredict the particle ratio , by a factor @xmath342 . regarding the data ratio itself
, it turns out the ratio first rises with @xmath7 , before seemingly saturating at a value of 0.15 , which could suggest that the hierarchy between and production becomes constant at high @xmath7 . between ( + ) and ( + ) as a function of transverse momentum , for collisions at @xmath0 = 7 .
the ratio is given for alice data as well as pythia perugia 2011 ( p2011 ) .
as a consistency check , the ratio of the tsallis functions fitted to the data or to the model predictions is also shown . ]
values ( @xmath36 ) for different species identified by the alice experiment in collisions , at @xmath0 = 0.9 @xcite and 7 .
these values are compared to the isr parametrisation ( performed from the @xmath37 , k , p species in collisions at ) and star measurements in central @xmath38 collisions at @xcite . ] the figure [ fig : maire - fig3-alicemeanpt - ppcollisions ] presents the alice preliminary values obtained for the mean @xmath7 ( @xmath36 ) of various identified particles , as measured in collisions at 7 . for comparison , the data are plotted together with results published by alice for data at @xmath0 = 0.9 @xcite as well as by star for the central @xmath38 collisions at @xmath39 = 0.2 @xcite . for a given particle studied in collisions
, it appears that the @xmath36 rises with @xmath0 . at ,
the data come to reach the values obtained in the most central collisions at rhic , raising the question of the physical similarities between these different systems at different energies .
with the results presented in this report , the alice experiment provides the first lhc measurement of and at central rapidity .
together with and results , it actually enables an extension of the excitation function in a system .
this is of importance to define the baseline necessary for the studies related to the so - called _ strangeness enhancement _
besides , although the tune perugia 2011 is a priori the current most suitable pythia tune for the description of hyperons in collisions at lhc energies , it is observed that it underestimates the data at intermediate @xmath7 . | in the perspective of comparisons between proton - proton and heavy - ion physics , understanding the production mechanisms ( soft and hard ) in that lead to strange particles is of importance .
measurements of charged multi - strange ( anti-)baryons ( and ) are presented for collisions at @xmath0 = 7 .
this report is based on results obtained by alice ( a large ion collider experiment ) from the 2010 data - taking .
+ taking advantage of the characteristic cascade - decay topology , the identification of , , and can be performed , over a wide range of momenta ( e.g. from 0.6 to 8.5 for , with the present statistics analysed ) .
the production at central rapidity ( @xmath1 ) as a function of transverse momentum , @xmath2 , is presented .
these results are compared to pythia perugia 2011 predictions . |
the galactic center ( gc ) is an exceptional region for testing massive star formation and evolution models .
it contains 10% of the present star formation activity in the galaxy , yet fills only a tiny fraction of a percent of the volume in the galactic disk500 pc and thickness of @xmath060 pc that is centered on the galactic nucleus and is coincident with a region of increased dust and gas density , often referred to as the `` central molecular zone '' @xcite . ] .
the initial conditions for star formation in the gc are unique in the galaxy .
the molecular clouds in the region are extraordinarly dense , under high thermal pressure , and are subject to a strong gravitational tidal field .
@xcite argue that these conditions may favor the preferential formation of high mass stars .
being the closest galactic nucleus , the gc gives us an opportunity to observe processes that potentially have wide applicability in other galaxies , both in their centers and in the interaction regions of merging galaxies .
finally , the gc may be the richest site of certain exotic processes and objects in the galaxy , i.e. runaway stellar mergers leading to intermediate mass black holes and stellar rejuvination through atmospheric stripping , to name a few .
this review is primarily concerned with massive star formation in the region . for thorough reviews on a variety of topics concerning the galactic center ,
see @xcite , @xcite , @xcite , and @xcite .
the star formation efficiency in the gc appears to be high . plotting the surface star formation rate ( @xmath15 yr@xmath2 pc@xmath3 ) versus surface gas density ( @xmath4400 pc@xmath3 ) in a `` schmidt plot '' suggests an efficiency of nearly 100% , comparable to that of the most intense infrared circumnuclear starbursts in other galaxies and a factor of twenty higher than in typical galaxies ( see figure 7 in * ? ? ?
it is also higher than that elsewhere in the galaxy ; commensurately , stars in the gc emit about 5 - 10% of the galaxy s ionizing radiation and infrared luminosity .
@xcite review the content and conditions of the interstellar medium in the `` central molecular zone '' ( cmz ) , noting that the molecular clouds in the region are extraordinarily dense ( @xmath5 ) and warm ( @xmath6 ) with respect to those found in the disk of the galaxy .
@xcite argue that the density and internal velocities of clouds in the gc are a direct result of the strong tidal fields in the region , i.e. only the dense survive .
@xcite argue that the inexorable inflow of molecular material from further out in the galaxy powers continuous and robust star formation activity in the region .
it is still unclear how magnetic field strength affects star formation .
if it does matter , then the gc might be expected to reveal such effects .
the strength of the magnetic field in the gc has been estimated through far infrared polarized light from aligned dust grains @xcite and zeeman splitting of the oh molecule @xcite . in both cases ,
the field is inferred to be of milliguass strength .
however , @xcite argue strongly that these strengths are localized to bundles that delineate the extraordinary non - thermal filaments in the region @xcite , and are not representative of the field strength that is pervasive in the region
. if this is correct , then the fields inside gc molecular clouds may not be so strong versus those inside disk clouds ( @xmath7 g ) .
metals in molecular clouds can provide cooling that aids protostellar collapse , but they also create opacity to the uv flux , winds , and bipolar outflows that emanate from newly formed stars . measurements of metallicty in the galactic center span a range of solar , observed in stars @xcite , to twice solar , observed in the gas phase @xcite , to four times solar , observed through x - ray emission near the very center @xcite .
the errors from the stellar measurements are the smallest and suggest that stars in the gc are formed from material with roughly solar abundances .
present - day star formation in the gc is somewhat subdued compared to the episodes that produced the massive clusters we now see
. a dozen or so ultra - compact hii regions are distributed throughout the central 50 pc , each containing one or a few o - stars still embedded in their natal environs . @xcite
identify most of these sources in radio continuum observations ( see figure [ radio ] ) . @xcite and @xcite
infer lyman - continuum fluxes that are comparable to that expected from a single o7v star in each of the h1-h5 and a - d uchii regions .
@xcite find that several of the recently formed stars in these regions have broken out of their dust shroud , revealing spectra of young massive stars ; see also @xcite and @xcite for additional examples .
0.7truein , title="fig:",height=377 ] a bit further from the gc , the sgr b2 molecular cloud harbors a massive star cluster in the making and is home to the most intense present - day star formation site in the galaxy @xcite . within the next few myr
, this activity should produce a star cluster that is comparable in mass to the arches cluster ( see figure [ mcgrath ] ) .
@xcite note evidence in support of a cloud - cloud collision as the origin for the intense star formation in sgr b2 ; these include velocity gradients , magnetic field morphology , shock - enhanced molecular emission , shock - induced molecular evaporation from dust grains , and distinctly different densities of certain molecular species throughout the cloud .
0.3truein o and oh masers overplotted on 7 mm contours for a small portion of the sgr b2 cloud .
the activity in this region is typical of that found near the fifty or so ultra - compact hii regions in sgr b2 .
[ mcgrath],title="fig:",width=264 ]
there is ample evidence for persistent star formation in the gc in the form of upper - tip asymptotic giant branch stars distributed throughout the region @xcite .
figure [ blum1 ] shows a plot for some of these stars , based on spectroscopic data , overlaid with intermediate age model isochrones @xcite .
note that the giants and supergiants in this plot require ages that span a few to a few .
one comes to similar conclusions by analyzing photometry of the field population in the gc .
@xcite use observed luminosity functions to determine that the star formation rate has been roughly constant for the lifetime of the galaxy in the gc , similar to the suggestion in @xcite based on the sharp increase in unresolved infrared light towards the center and a mass - budget argument .
figure [ sfpanels ] shows model and observed luminosity functions ( _ right _ ) for various star formation scenarios ( _ left _ ) over the lifetime of the galaxy , assuming a salpeter imf @xcite for masses above 10 , and a flat slope below this mass .
the observations were obtained with hst / nicmos and have been corrected for incompleteness .
the `` burst '' models ( panels 1 , 2 , 4 , and 5 ) produce unrealistic ratios of bright to faint stars in the luminosity functions , especially for the red clump near a dereddened k - band magnitude of 12 . the continuous star formation model ( panel 3 ) best fits the data . 0.3truein 15 - 25 , whereas fainter stars are descendant from lower mass main sequence stars having a few to 15 .
the presence of these stars in the gc demonstrates intermediate age star formation of massive stars.[blum1],title="fig:",width=226 ] 0.3truein ) in stars within the region .
the observations have been corrected for incompleteness .
the third panels from the top , i.e. continuous star formation , best fit the data . the observed turn - down at the faint end appears to be real and is only well fit only by assuming a very high lower mass turnover.[sfpanels],title="fig:",width=377 ]
the majority of recent star formation activity in the gc over the past 10 myr produced three massive clusters : the central cluster , the arches cluster , and the quintuplet cluster .
the following sections describe the stellar content in the clusters and the resultant implications for star formation in the region .
they closely follow recent reviews @xcite , with updates , as summarized in table [ proptable ] .
.properties of massive clusters in the galactic center[proptable ] [ cols= " < , > , > , > , > , > , > , > , > " , ] oddly , the increased density of the young stars in the central arcsecond is not matched by the density distribution of old stars .
indeed , there is a curious absence of late - type stars in the central few arcseconds , as evidenced by a lack of stars with strong co absorption in their k - band spectra @xcite .
this dearth of old stars represents a true `` hole '' in three dimensional space , and not just a projection effect .
even the late - type stars that are projected on to the central parsec generally have relatively low velocities , suggesting dynamical evidence that the region nearest to the black hole lacks old stars @xcite .
the existence of such massive and young stars in the central arcseconds is puzzling , although it is perhaps only an extension of the original problem in understanding the origins of the young stars identified in the central parsec over 20 years ago .
table [ stable ] gives a list of recent papers regarding the origin of the `` s '' stars .
while there are over 30 papers listed in this table , they can be reduced to a few basic ideas .
one class of ideas considers the `` s '' stars as truly young . in this case , the `` origin '' of the `` s '' stars is often reduced to the case of massive star formation in the galactic center region and transportation of the products to the central arcsecond .
the other class regards the `` s '' stars as old stars that only appear to be young , i.e. via atmospheric stripping , merging , or heating .
both classes require new mechanisms that would be unique to the gc , and they both have considerable weaknesses .
for example , @xcite argue that stripped red giants would not be as bright as the `` s '' stars ( see * ? ? ? * for detailed confirmation ) .
see @xcite for a more thorough discussion of the strengths and weaknesses of these ideas .
if the `` s '' stars are truly young , then that would require massive clumps to form ob stars ( @xmath820 ) .
in addition , the clumps would have to form from very high density material in order for them to be stable against tidal disruption .
assuming that the stars formed as far away from the supermassive black hole as possible , while still permitting dynamical friction to transport them into the central arcsecond during their lifetimes , then the required densities must be @xmath910@xmath10 @xcite .
the average molecular cloud density in the gc is about five orders of magnitude less , so highly compressive events might be required to achieve the necessary densities .
alternatively , the required densities can be reduced if the stars are gravitationally bound to significant mass , i.e. a surrounding stellar cluster .
indeed , @xcite , @xcite , and @xcite showed that particularly massive clusters could form tens of parsecs outside of the center and be delivered into the central parsec in just a few million years .
the efficiency of this method is improved with the presence of an intermediate black hole in the cluster @xcite .
it is key in any of these cluster transport models that the host system have extremely high densities of @xmath910@xmath11 pc@xmath12 , comparable to the highest estimated central density of the arches cluster after core collapse @xcite .
detailed n - body simulations suggest that while these ideas may be relevant for the origins of the young stars in the central parsec , it is unlikely that they could explain the existence of the `` s '' stars in the central arcsecond .
there are relatively few clusters in the galaxy with as many massive stars as in the gc clusters .
ngc3603 has about a factor of two less mass than each of the gc clusters @xcite ; whereas , w1 has at least a factor of two greater mass @xcite .
the next nearest similarly massive cluster is r136 in the lmc @xcite .
all of these clusters , and the gc clusters , appear to have imf slopes that are consistent with the salpter value ( or slightly flatter ) and are young enough to still possess a significant massive star population .
it is remarkable to note that these massive clusters appear quite similar in stellar content , whether in the galactic disk , the gc , or even the lower metallicity environment of the lmc .
evidently , the star formation processes , and natal environments , that gave birth to these clusters must be similar enough to produce clusters that are virtually indistinguishable .
there are probably more massive clusters yet to be found in the galaxy .
the limited sample of known massive clusters is a direct result of extinction , as most star formation sites in the galaxy are obscurred by dust at optical wavelengths .
while infrared observations have been available for over 30 years , they have not provided the necessary spatial resolution , nor survey coverage , needed to probe the galactic disk for massive clusters .
recently , a number of groups have begun identifying candidate massive star clusters using near - infrared surveys with arcsecond resolution @xcite .
indeed , these surveys have already yielded a cluster with approximate initial mass of 20,000 to 40,000 @xcite , and one would expect more to be discovered from them .
the present - day sites of massive star formation in the galaxy have been known for some time through radio and far - infrared observations , as their hottest members ionize and heat nearby gas in molecular clouds . as one of many examples ,
consider w49 which is the next most massive star formation site in the galaxy compared to sgr b2 @xcite , and wherein the star formation appears to be progressing in stages over timescales that far exceed the individual collapse times for massive star progenitors .
this suggests a stimulus that triggers the star formation , perhaps provided by a `` daisy chain '' effect in which newly formed stars trigger collapse in nearby parts of the cloud .
similar suggestions are proposed in the 30 dor region surrounding r136 @xcite . while there is no evidence for an age - dispersed population in sgr b2 , @xcite suggest that the cloud was triggered to form stars through a cloud - cloud interaction .
massive star formation in the gc has produced an extraordinary sample of stars populating the initial mass function up to a cutoff of approximately 150 .
the ranges of inferred masses and observed spectral types are as expected from stellar evolution models , and the extraordinary distribution of stars in the region is a direct consequence of the large amount of mass that has fed star formation in the gc .
the origin of the massive stars in the central parsec , and especially the central arcsecond , remains unresolved .
i thank the following individuals for discussions related to this work : mark morris , bob blum , reinhard genzel , paco najarro , sungsoo kim , and peter tuthill .
tal alexander made substantial contributions to table 3 .
the material in this paper is based upon work supported by nasa under award no .
nng05-gc37 g , through the _ long term space astrophysics _ program .
beloborodov , a. m. , levin , y. , eisenhauer , f. , genzel , r. , paumard , t. , gillessen , s. , & ott , t. 2006 clockwise stellar disk and the dark mass in the galactic center .
arxiv astrophysics e - prints * * , arxiv : astro - ph/0601273 chiar , j. e. , adamson , a. j. , whittet , d. c. b. , & pendleton , y. j. 2003 spectroscopy of hydrocarbon grains toward the galactic center and quintuplet cluster .
astronomische nachrichten supplement * 324 * , 109 crowther , p.a . , hadfield , l. j. , clark , j. s. , negueruela , i. , & vacca , w. d. 2006 a census of the wolf - rayet content in westerlund 1 from near - infrared imaging and spectroscopy .
arxiv astrophysics e - prints * * , arxiv : astro - ph/0608356 eckart , a. , schdel , r. , & straubmeier , c. 2005 the black hole at the center of the milky way .
andreas eckart , rainer schdel , christian straubmeier .
london : imperial college press , isbn 1 - 86094 - 567 - 8 , 2005 , xxii+284 figer , d. f. , becklin , e. e. , mclean , i. s. , & morris , m. 1994 discovery of luminous nir sources associated with ionized gas near the galactic center .
assl vol . 190 : astronomy with arrays , the next generation * * , 545 figer , d. f. , morris , m. , geballe , t. r. , rich , r. m. , serabyn , e. , mclean , i. s. , puetter , r. c. , & yahil , a. 1999c high - resolution infrared imaging and spectroscopy of the pistol nebula : evidence for ejection .
, 759 groh , j. h. , damineli , a. , teodoro , m. , & barbosa , c. l. 2006 detection of additional wolf - rayet stars in the starburst cluster westerlund 1 with soar .
arxiv astrophysics e - prints * * , arxiv : astro - ph/0606498 harris , a. i. , krenz , t. , genzel , r. , krabbe , a. , lutz , d. , politsch , a. , townes , c. h. , & geballe , t. r. 1994 spectroscopy of the galactic center arches region : evidence for massive star formation .
nato asic proc . 445 : the nuclei of normal galaxies : lessons from the galactic center * * , 223 lang , c. c. , goss , w. m. , & wood , d. o. s. 1997 vla h92 alpha and h115 beta recombination line observations of the galactic center h ii regions : the sickle ( g0.18 - 0.04 ) and the pistol ( g0.15 - 0.05 ) . ,
275 lang , c. c. , johnson , k. e. , goss , w. m. , & rodrguez , l. f. 2005 stellar winds and embedded star formation in the galactic center quintuplet and arches clusters : multifrequency radio observations .
, 2185 lindqvist , m. , habing , h. j. , & winnberg , a. 1992 oh / ir stars close to the galactic centre .
ii - their spatial and kinematic properties and the mass distribution within 5 - 100 pc from the galactic centre . , 118 mason , b. d. , gies , d. r. , hartkopf , w. i. , bagnuolo , w. g. , jr . , ten brummelaar , t. , & mcalister , h. a. 1998 iccd speckle observations of binary stars .
xix - an astrometric / spectroscopic survey of o stars .
, 821 nayakshin , s. , & sunyaev , r. 2005 the ` missing ' young stellar objects in the central parsec of the galaxy : evidence for star formation in a massive accretion disc and a top - heavy initial mass function .
, l23 nelan , e. p. , walborn , n. r. , wallace , d. j. , moffat , a. f. j. , makidon , r. b. , gies , d. r. , & panagia , n. 2004 resolving ob systems in the carina nebula with the hubble space telescope fine guidance sensor .
, 323 sjouwerman , l. o. , habing , h. j. , lindqvist , m. , van langevelde , h. j. , & winnberg , a. 1999 oh / ir stars as signposts for ancient starburst activity in the galactic center .
asp conf .
ser . 186 : the central parsecs of the galaxy * 186 * , 379 wang , q. d. , dong , h. , & lang , c. 2006 the interplay between star formation and the nuclear environment of our galaxy : deep x - ray observations of the galactic center arches and quintuplet clusters .
arxiv astrophysics e - prints * * , arxiv : astro - ph/0606282 yusef - zadeh , f. , nord , m. , wardle , m. , law , c. , lang , c. , & lazio , t. j. w. 2003 nonthermal emission from the arches cluster ( g0.121 + 0.017 ) and the origin of gamma - ray emission from 3eg j1746 - 2851 .
, l103 zhao , j .- h . ,
desai , k. , goss , w. m. , & yusef - zadeh , f. 1993 vla radio recombination line observations of ionized gas in the -30 kilometers per second molecular cloud ( g0.04 + 0.03 ) near the galactic center . i. the discrete radio sources . , 235 | the galactic center is a hotbed of star formation activity , containing the most massive star formation site and three of the most massive young star clusters in the galaxy .
given such a rich environment , it contains more stars with initial masses above 100 than anywhere else in the galaxy .
this review concerns the young stellar population in the galactic center , as it relates to massive star formation in the region .
the sample includes stars in the three massive stellar clusters , the population of younger stars in the present sites of star formation , the stars surrounding the central black hole , and the bulk of the stars in the field population .
the fossil record in the galactic center suggests that the recently formed massive stars there are present - day examples of similar populations that must have been formed through star formation episodes stretching back to the time period when the galaxy was forming . |
quantum walks ( qws ) provide the natural - yet unitary - extension of the classical random walk ( rw ) to the quantum regime @xcite , where the superposition principle allows a wave function to explore different paths simultaneously . the resulting interference in the transition amplitudes is cause to effects like the linear increase in time of the mean travelled distance of the quantum walker @xcite .
this ballistic behavior contrasts with the diffusive behavior of a classical particle performing a rw , whose mean travelled distance increases with the square root of time .
single - particle qws , as realized in various experiments @xcite , can also be mimicked with classical wave mechanics @xcite .
in contrast , qws with two particles @xcite show purely quantum features due to many - particle interference or entanglement that no classical system can provide . for larger particle numbers ,
it is unknown how the emerging hierarchy of many - particle interference effects in terms of the number of contributing particles affects physical observables .
we will derive a general expression for multi - mode particle number correlation functions which gives access to this hierarchy , and provide first examples as manifest in prominent interference structures in two - mode number correlation functions .
while the qw constitutes a paradigmatic testing ground for the study of such many - particle interference , the formalism we present applies to _ any _ linear , non - interacting scattering scenario . in a many - particle qw ,
the distinction between effects of single - particle interference and those due to genuine many - particle interference is crucial .
since distinguishable particles show no many - particle interference , a scenario with non - interacting , distinguishable particles @xcite is equivalent to many subsequent , independent single - particle walks . given the position probability distribution of an individual particle , standard combinatorics suffices to derive the many - body counting statistics .
therefore , the differences in the behavior of distinguishable and indistinguishable particles highlight true multi - particle interference .
we will see that these differences are not detectable by the mean particle number , but only by observables which capture many - particle correlations .
the counting statistics thus encodes an unambiguous signature of many - particle coherence and interference and is readily accessible in the experiment .
we consider a many - particle qw @xcite in one dimension , physically realized by n particles which propagate through an array of l unbiased beam splitters ( bs ) as depicted in fig .
[ arraypic ] @xcite .
the transverse position of the beam splitters in the array defines the particle position , and subsequent bs rows represent the discrete steps in time . ) .
particles start in modes with label @xmath0 , propagate from top to bottom through the adjacent rows of bs , and are eventually detected in modes with label ( n ) , on exit .
, width=264 ] a particle that passes a single bs will leave it in a coherent superposition of the two output modes , as described by the unitary @xmath1 matrix @xmath2 while the particles propagate through the array , we have to distinguish two inequivalent steps , as depicted in fig . [ arraypic ] : the direct sum @xmath3 of the single bs matrices u describes odd time steps , where modes 1 and 2 enter one beam splitter , modes 3 and 4 enter the next and so forth . in the following step , modes 2 and 3 couple into the same bs , and so do modes 4 and 5 etc .
even steps are thus described by @xmath4 , with matrix elements @xmath5}$ ] of @xmath6 .
the @xmath7 matrix @xmath8 describes the evolution for n time steps .
the creation operator of a particle @xmath9 in mode j at time step 0 is mapped onto a superposition of the creation operators at time step n , with 2l the maximum number of modes : @xmath10 given an n - particle initial state , distributed over 2l input modes , @xmath11 eq . ( [ unitev ] ) implies the evolution into the output state @xmath12 in ( [ inistate ] , [ prodfin ] ) , @xmath13 and @xmath14 are the vacuum state and the occupation number of mode j ( such that @xmath15 ) , respectively .
note that the number of terms in ( [ prodfin ] ) grows exponentially with n , so that simulations of the dynamics of many - particle qws are practically impossible for larger systems .
however , expectation values of many experimental observables can be derived without complete knowledge of the full final state .
one class of such observables which reveals many - body interference are particle number correlation functions @xmath16 of m output modes @xmath17 . for m=1 ,
this is the single - mode particle expectation number , @xmath18 , simply given by the sum of 2l single - particle densities of a single - particle qw , @xmath19 where @xmath20 is the transition amplitude from mode k to mode i , and we left out the dependence of @xmath20 on n , for simplicity of notation .
( [ density ] ) holds for both , fermions and bosons , as well as for distinguishable particles . within these different particle - families ,
various distinct initial states are mapped on final states with the same mean particle number by the beam splitter array .
many - particle interference does not manifest on this level , and no information about the many - particle coherence of the state is obtained , as anticipated above .
this is different when we correlate detection events in two different modes , i.e. m=2 ( the upper / lower @xmath21-sign refers to bosons / fermions ) : @xmath22 the two sums in this equation represent two physically distinct processes : in the first , the particles originate from _ different _ modes .
the two - particle amplitude @xmath23 describes the simultaneous transition of one particle from mode k to i , and of another one from l to j. due to the indistinguishability of the particles , this amplitude interferes with that where k and l are exchanged .
the second sum accounts for particles that originate from the _ same _ mode .
the corresponding many - particle amplitude @xmath24 is invariant upon exchange of the input modes , and hence exhibits no many - particle interference .
( [ density ] , [ 2correlator ] ) exhaust all possible correlations between two particles injected into the bs array . only higher order correlations between a larger number of particles can exhibit genuine many - particle interference effects . since the number of distinct correlations increases like the factorial of the particle number n , a general expression for the m - mode particle number correlation function is desirable , and can be shown to read @xmath25 where @xmath26 . the function @xmath27 is given by unity for bosons ( b ) , and by the sign of the permutation in its argument for fermions ( f ) .
the outer , incoherent sum in eq .
( [ expval ] ) runs over all ordered combinations @xmath28 of input modes the particles may have originated from .
every single summand represents a physically distinct event .
each such event , in turn , is given by the _
coherent _ sum over all s(@xmath29 ) permutations @xmath30 of @xmath29 , due to the particles indistinguishability , where @xmath31 represents the j - th element of @xmath30 .
clearly , ( [ expval ] ) reproduces ( [ density ] ) and ( [ 2correlator ] ) for m=1 and m=2 , respectively , with the two sums in ( [ 2correlator ] ) corresponding to @xmath32 and @xmath33 in ( [ expval ] ) .
expression ( [ expval ] ) permits to infer a large variety of many - particle observables , including the particle number counting statistics . since the latter are restricted to occupation numbers zero or one for fermions , we concentrate on bosons in the following .
the counting statistics of mode i , i.e. the probability @xmath34 to find k particles in mode i , is defined by the linear relation @xmath35 where the @xmath36 are determined via eq .
( [ expval ] ) . this relation can be inverted to yield @xmath34 , if all moments @xmath37 are known .
analogously , the counting statistics @xmath38 of two modes i and j , _ i.e. _ the joint probability to find exactly @xmath39 particles in mode i and @xmath40 particles in mode j , is given in terms of @xmath41 with @xmath42 .
thus all two - mode moments @xmath43 are required for the two - mode counting statistics .
multi - mode counting statistics for more than two modes are obtained in strictly analogous fashion . with these tools
, we can now fully characterize the distribution of particles among the modes , and work out the observable consequences of many - particle interference : we start out with n=8 particles , launched in eight adjacent modes at the center of an array of length 2l=50 , and initially in a product state ( [ inistate ] ) .
[ onesite ] shows the probability for mode 25 to be occupied by k particles after @xmath44 steps .
( k ) for mode i=25 after @xmath44 steps on a logarithmic scale .
bosonic bunching enhances p@xmath45(k ) for bosons by several orders of magnitude ( blue circles ) as compared to the result for distinguishable particles ( red squares ) .
the inset shows the distribution for small k.,width=283 ] the probability decreases with k , both for bosons and distinguishable particles .
since , however , for bosons all many - particle amplitudes that contribute to the coherent sum in ( [ expval ] ) are identical , they interfere constructively .
consequently , the probability to find more than two particles in one mode decreases much slower for bosons than for distinguishable particles . for the latter , p@xmath45(k )
exhibits poissonian decrease like 1/k!. in contrast , in the limit of many particles , p@xmath45(k ) can be shown to approach exponential decrease in the bosonic case , which is apparent in fig .
[ onesite ] for low and intermediate values of k. for k = n=8 , p@xmath45(k ) is equal to @xmath46 in ( [ qone ] ) .
there , all n !
amplitudes in the coherent sum in ( [ expval ] ) are necessarily equal .
thus , bosonic bunching @xcite induces a factor n!=8 ! as compared to distinguishable particles . for @xmath47 , p@xmath45(k )
is a sum of several @xmath48 , what prevents such simple derivation of the bosonic value of p@xmath45(k ) .
further note that , since the number of constructively interfering amplitudes grows monotonically with the number of detected particles , the distribution in fig .
[ onesite ] is smooth ( in contrast to the strong fluctuations in fig .
[ twomodecut ] below ) , and , in particular , it is qualitatively independent of the considered output mode .
in contrast to the above single - mode counting statistics , _ distinct _ many - particle amplitudes contribute to the coherent sum in ( [ expval ] ) when _ several _ output modes are correlated , and give rise to characteristic many - particle interference effects beyond the smooth bunching observed in fig .
[ onesite ] . for
these interference effects to fully develop , we need to propagate the many - particle state over a larger number of steps ( n=20 ) , such that all single particle wave functions spread over the entire lattice , and mode - to - mode correlations can build up .
our figure of merit is then the conditional probability @xmath49 to find @xmath50 particles in mode @xmath51 , with a total of @xmath52 particles in these modes . indeed , as evident in fig .
[ twomodecut ] , many - particle interference of indistinguishable particles has a marked impact on @xmath53 as compared to the case of distinguishable ones .
while the latter always implies a single - peaked dependence of @xmath53 on the particle imbalance @xmath54 , many - particle interference manifests in strong modulations of @xmath53 for bosons , with pronounced maxima for even values of @xmath55 , i.e. for odd particle numbers in either one of both modes ( remember that n=8 ) .
furthermore , the many - particle interference condition is strongly affected by the specific choice of the output modes i and j , as obvious from a comparison of figs .
[ twomodecut]a ) and b ) : an asymmetric choice of i and j with respect to the lattice center induces a strong asymmetry in @xmath56 . what s more , the interference signal sensitively depends on the total number of particles
n in the system , in a non - monotonic way , as apparent from fig .
[ twomodecut ] c ) , where the bosonic signal is shown for different n. this allows for a clear distinction of multi - particle interference contributions of different order .
when not all particles contribute to the signal in modes i and j , i.e. for @xmath57 , many of the events which contribute to @xmath58 can be distinguished by the final positions of the remaining @xmath59 particles and thus can not interfere anymore , what leads to a smoother dependence of @xmath58 on @xmath60 .
this is clearly apparent in figs .
[ twomodecut ] a),b ) , as well as from ( [ expval ] ) : for expectation values that involve the maximum number @xmath61 of creation / annihilation operators , the incoherent sum collapses to a single term while the coherent sum contains up to n
! terms . for @xmath62 ,
many more incoherently added terms contribute , and , thus , interference effects are averaged out . among two modes after n=20 steps of n=8 particles , for lattice size 2l=50 , conditioned on finding m=4 or 8 particles .
a ) joint probability distribution of the symmetric ( with respect to the lattice center ) modes i=19 , j=32 to find m=4 particles ( dashed ; distinguishable : red squares , bosons : blue circles ) and m=8 particles ( solid ; distinguishable : red squares , bosons : blue circles ) .
b ) same as a ) , for asymmetric modes i=18 , j=32 .
c ) @xmath63 of modes i=19 ,
j=32 for n=4 ( orange diamonds ) , n=6 ( green squares ) and n=8 ( black circles).,width=302 ] let us finally stress that the observed many - particle interference effects even persist after an average of @xmath53 over all pairs @xmath64 : fig .
[ twomodecutmean ] shows that the bosonic signal is markedly different from the classical result , after a given number of steps as well as in its dependence on n. while the two - mode correlation is single - peaked and essentially n-_independent _ for distinguishable particles , some of the interference structure of fig .
[ twomodecut ] prevails under the average , and , furthermore , shows considerable variation with n. this latter dependence is , similarly to the case considered in fig .
[ twomodecut ] , reduced for @xmath65 . , averaged over all pairs of modes , after a ) n=19 steps and b ) n=20 steps .
color code as in fig .
[ twomodecut ] .
note that while the distribution is essentially invariant under changes of n for distinguishable particles , the correlations fluctuate with n , in the bosonic case.,width=302 ]
in summary , many - particle interferences between n indistinguishable particles injected into the 2l input modes of a beam splitter array give rise to strong bunching effects in the single mode counting statistics , as well as to a robust , pronounced and non - monotonic dependence of the two - mode correlation function on total particle number and particle imbalance between both modes .
the many - particle interference correction induces large deviations from the classical ( distinguishable particle ) result .
this allows to characterise the witnessed interference phenomenon in terms of the number of contributing particles - an information unaccessible , e.g. , in few - particle quantum walks .
this is of particular interest , e.g. , in the many - particle transport across disordered media @xcite , which is also covered by our central result , eq .
( [ expval ] ) , that actually allows to explore the full abundance of statistical observables of any non - interacting system with an arbitrary number of particles .
the authors acknowledge fruitful and stimulating discussions with p. m. dibwe and f. petruccione . partial financial support by the german federal ministry for science and education under the grant sua 08/008 is gratefully acknowledged .
m.c.t . acknowledges financial support by studienstiftung des deutschen volkes . | we study quantum walks of many non - interacting particles on a beam splitter array , as a paradigmatic testing ground for the competition of single- and many - particle interference in a multi - mode system .
we derive a general expression for multi - mode particle - number correlation functions , valid for bosons and fermions , and infer pronounced signatures of many - particle interferences in the counting statistics . |
is the electroweak phase transition accompanied by the condensation of some defects ? what is happening with the defects at the crossover region
? is there a range of parameters supporting a dilute gas of such defects in the broken phase and how are they related to the well - known sphaleron configurations ?
these questions have been explored during the last year within the @xmath1 higgs model .
the @xmath0-vortex string @xcite corresponds to the abrikosov
olesen @xcite vortex solution related to the abelian subgroup of @xmath1 embedded into the electroweak theory @xcite .
@xmath0-strings have been demonstrated to be unstable both at zero and at finite temperatures @xcite .
sphalerons @xcite are unstable configurations as well , being transition states ( saddle points ) between vacua with different chern - simons number .
classical sphalerons have been constructed on the lattice in refs .
@xcite by a specific saddle point cooling technique ( these configurations have been important for our project ) .
baryon asymmetry should not be washed out by sphaleron processes after the electroweak symmetry breaking has been completed .
this requirement puts stringent conditions on the strength of the phase transition because the activation energy of these barriers strongly depends on the strength of the electroweak phase transition @xcite .
recently , lattice studies have shown ( see @xcite for a conclusive summary ) that neither the simplified @xmath1 higgs model nor the @xmath2 model have a thermal transition of the necessary strength , given the experimental bounds on the higgs mass .
this observation has triggered our interest to investigate whether , for higgs masses below the end of the phase transition @xcite and in the crossover region above , topological excitations might be excited with a non - negligible statistical weight .
these historical circumstances might explain why defects in the context of the electroweak phase transition are a rather new topic within lattice gauge theory . in this contribution
we briefly describe our lattice definitions and present selected results , some of them having not been published before .
details can be found in refs .
our study has been performed within the dimensionally reduced @xmath1 higgs theory which provides an effective representation of the electroweak theory in some range of higgs masses and at temperatures near to the electroweak scale .
for the formulation of the lattice version of the effective theory see the talk of a. schiller at this conference @xcite .
the lattice gauge coupling @xmath3 ( with @xmath4 ) gives the lattice spacing in units of temperature , and the hopping parameter @xmath5 substitutes @xmath6 driving the transition .
the parameter of the phase transition or the crossover can be translated into temperature @xmath7 and a higgs mass @xmath8 ( see _ e.g. _ ref .
@xcite and references therein ) .
notice that the higgs field is usually written as a multiple of a @xmath1 matrix @xmath9 , here we need it in the form of a @xmath10-component isospinor @xmath11 .
the gauge invariant and quantized lattice definition @xcite of elementary @xmath0-vortices and nambu monopoles is closely related to the definition in the continuum theory @xcite .
first we define a composite adjoint unit vector field @xmath12 @xmath13 are pauli matrices . for the following purpose
, @xmath14 plays a role similar to the adjoint higgs field in the definition of the t hooft polyakov monopole @xcite within the georgi glashow model .
it helps to define the gauge invariant flux @xmath15 through the plaquette @xmath16 , @xmath17\bigr ) \label{ap}\ ] ] via the projected links @xmath18 the auxiliary higgs field @xmath14 is covariantly constant with respect to these @xmath1-valued links .
originally , nambu monopoles are topological defects of an abelian field which is defined in the unitary gauge ( @xmath19 and @xmath20 ) by the phases @xmath21 .
this field behaves as a compact abelian field with respect to residual abelian gauge transformations which leave the unitary gauge condition intact .
the gauge invariant flux would be extracted from its plaquettes @xmath22 as usual : @xmath23 .
the magnetic charge of nambu monopole cubes and the @xmath0-vorticity number of plaquettes can alternatively be defined in a gauge invariant way @xcite referring to ( [ ap ] ) .
the monopole charge @xmath24 inside the cube @xmath25 is defined in terms of the gauge invariant fluxes ( [ ap ] ) penetrating the surface @xmath26 : @xmath27 the @xmath0vorticity number of the plaquette @xmath28 is defined @xcite as follows : @xmath29 where @xmath15 has been given in ( [ ap ] ) , and @xmath30 is a plaquette variable written in terms of the abelian links @xmath31 .
a @xmath0vortex is formed by links @xmath32 of the dual lattice which are dual to plaquettes @xmath16 which carry a non - zero vortex number ( [ sigman ] ) : @xmath33 .
@xmath0vortex trajectories are either closed or begin / end on nambu ( anti- ) monopoles : @xmath34 this is how we actually localize nambu monopoles .
when the embedded defects are physical excitations with some intrinsic size , one should be able to describe them on finer lattices as _ extended _ topological objects .
an extended monopole ( vortex ) of physical size @xmath35 is defined on @xmath36 cubes ( @xmath37 plaquettes , respectively ) @xcite .
the charge of monopoles @xmath38 on bigger @xmath36 cubes @xmath39 is constructed analogously to that of the elementary monopole ( [ jn ] ) with elementary @xmath40 plaquettes in terms of @xmath41 replaced by @xmath42 wilson loops ( extended plaquettes ) . in pure gauge theory , within the maximally abelian gauge , these objects are called type - i extended abelian monopoles we have numerically investigated the behavior of the type - i extended vortices of some _ fixed _ physical size on a series of finer and finer lattices towards the continuum limit .
our results suggest that the continuum limit for type - i extended vortices may exist .
we remark that an alternative , type - ii , construction @xcite of extended objects fails to show a physically sensible behavior towards continuum limits .
we have scanned the phase transition with elementary defects at @xmath43 gev ( strong first order ) and 70 gev ( weak first order ) for @xmath44 on a relatively small lattice @xmath45 . at the lower higgs mass we observed a discontinuity of the densities @xmath46 of nambu monopoles and @xmath47 of vortices .
simultaneously we have measured the percolation probability @xmath48 derived from the cluster correlation function @xcite @xmath49 where only those connected clusters @xmath50 of vortex lines ( @xmath51 labels distinct vortex clusters ) contribute to the nominator which pass through the points @xmath52 and @xmath53 . connected clusters @xmath50
are called percolating clusters if they contribute to the limit @xmath54 .
the percolation probability has a finite jump to zero at the pseudocritical @xmath55 for @xmath43 gev .
the same study at @xmath56 gev , already near to the endpoint of the first order transition , is summarized in figure 1 and shows the percolation probability continuously vanishing towards @xmath57 .
this is an inflection point of the densities @xmath58 and @xmath59 where they are approximately equal to half of their values in the symmetric phase .
for @xmath60 the densities decrease exponentially . for the case
@xmath43 gev ( where we find zero densities of elementary defects on the broken side ) we have made some tests whether the densities of extended defects on the symmetric side admit a continuum limit .
we have chosen elementary vortices and monopoles on a lattice simulated at @xmath61 and compared them with extended defects of size @xmath62 for simulations at @xmath63 and @xmath64 , respectively .
a lattice size @xmath45 corresponding to a shrinking volume has been taken for all three cases because we did not expect severe finite size effects at such a strong transition .
this ( preliminary ) test is shown in figure 2 in the form of histograms of average vortex and monopole densities per configuration .
the measurements have been performed near to the respective @xmath65 for the three values of @xmath66 and gave always a two - state signal with a density distribution peaking at or near to zero on the symmetric side .
there is no way to perform an additive renormalization of the density at the broken side by taking differences between the phases ( in analogy to the higgs modulus squared @xmath67 , ref .
@xcite ) . with a box size @xmath68 ( @xmath69 )
we seem to be at the upper edge of the size distribution of physical vortices . with increasing @xmath66 , our construction for extended defects seems to describe the vorticity better .
this might explain the monotonous increase to some ( still unknown ) _ continuum _ density of vortices / monopoles .
physically the crossover region is more interesting since we can study the properties of isolated defects in a not - so - dilute environment .
we show in figure 3 the behavior of densities and percolation probability @xmath54 _ vs. _
@xmath5 for @xmath70 gev at @xmath44 for elementary defects .
we were wondering whether there is an universal percolation temperature for vortices of extendedness @xmath35 at respective @xmath71 . in order to prevent bigger finite size effects
, now we kept the physical volume constant using @xmath72 lattices .
the percolation probability @xmath54 and the densities are shown in figure 4 _ vs. _ @xmath5 as measured for @xmath73 on a @xmath74 lattice . from all critical @xmath75 values
we estimate @xmath76 or 130 gev ( without or with @xmath77-quarks ) .
this corresponds to a higgs mass @xmath78 or 103 gev ( _ cf .
_ @xmath70 gev ) .
the cluster structure of the configurations changes drastically at this temperature .
we illustrate this for @xmath73 vortices in figure 5 with the average number of clusters and the average length per cluster _ vs. _ @xmath5 .
our results obtained in an effective @xmath79 equilibrium approach suggest a scenario with a ( few ) percolating clusters decaying into smaller ones , vortex rings and monopoliums , a process to be described by a kinetic approach to the crossover .
note that the average length per cluster has a long tail at a level of two .
we propose to interpret the half density @xmath80 as the density of spalerons on the broken side of the crossover .
we support this conjecture by what we have found @xcite to be the signature of a _ classical _ lattice sphaleron with respect to the new ( @xmath0-vortex and nambu monopole ) degrees of freedom .
figure 6 shows one of the solutions of ref .
@xcite with a nambu monopole - antimonopole pair in the center , compared with a view in the maximally abelian gauge .
c + + one might argue that the defects counted by the lattice operators ( [ jn ] ) and ( [ sigman ] ) might be mere artifacts having nothing in common with the vortex solutions in the continuum .
figure 7 shows that this is not the case . here
the average gauge field action per plaquette and the higgs field modulus squared averaged over the corners of a plaquette is shown as function of @xmath5 , separately for plaquettes with @xmath81 and @xmath82 .
the data refer to @xmath83 gev and @xmath84 .
one can clearly see that the modulus of the higgs field is smaller near to the vortex trajectory than outside the vortex for all values of the coupling @xmath5 . moreover , the value of the gauge field energy near to the vortex center is larger than in the bulk ( the increasing error of the gauge field action inside the vortex at the largest @xmath5 reflects the small number of plaquettes with non - vanishing vorticity ) .
these results show that @xmath0vortices are physical objects which resemble the characteristic features of the classical vortex solutions in continuum : the vortices carry the excess of the gauge field energy and the higgs modulus decreases near to the vortex center .
in future we have to complete these studies by a more systematic exploration of the continuum behavior / size distribution of defects .
furthermore we plan to extend our considerations to more realistic models with non - zero weinberg angle .
last but not least , euclidean simulations seem to be necessary to clarify the connection of the dynamics of vortices with the evolution of the chern - simons number .
this may shed some light on the baryogenesis scenario @xcite based on the dynamical properties of the @xmath0vortex string network .
the authors have benefited from discussions with l. mclerran , m. mller - preussker , m. i. polikarpov , v. a. rubakov , k. rummukainen , k. selivanov , t. suzuki and p. g. tinyakov .
we are very grateful to pierre van baal and margarita garcia
perez for providing the sphaleron lattice configurations .
a. v. pochinsky , m. i. polikarpov and b. n. yurchenko , phys .
a154 ( 1991 ) 194 ; t. l. ivanenko , a. v. pochinskii and m. i. polikarpov , phys .
b302 ( 1993 ) 458 ; a. hulsebos , nucl .
phys . b ( proc .
suppl . ) 34 ( 1994 ) 695 .
t. vachaspati and g. b. field , phys .
lett . 73 ( 1994 ) 373 ; ibid . 74 ( 1995 ) ( errata ) ;
t. vachaspati , in the proceedings of the nato workshop on `` electroweak physics and the early universe '' , sintra , portugal ( 1994 ) ; series b : physics vol .
338 , plenum press , new york ( 1994 ) , e - print archive : hep - ph/9405286 . | new defects ( @xmath0-vortices and nambu monopoles ) are found to become thermodynamically relevant for the broken phase near to the ( weakly first order ) electroweak phase transition , and below the crossover for higher higgs mass . the symmetric phase is characterized by vortex condensation ( percolation ) .
the percolation transition persists in the crossover region .
the quasiclassical nature of the vortices is supported by correlations with higgs field and gauge field energy .
sphalerons are shown to be related to monopolium bound states . |
in this paper , we study behavior of principal curvatures of wave fronts in the euclidean @xmath1-space @xmath2 with non - degenerate singular points which contain cuspidal edges and swallowtails , and their applications .
principal curvatures for regular surfaces play important roles to study the ( extrinsic ) differential geometry of surfaces and related topics .
for instance , types of singularities of parallel surfaces and focal surfaces of regular surfaces are closely related to critical points of principal curvatures , such points are called ridge points . in fact , porteous @xcite showed relations between cuspidal edges on a focal surface and ridge points on an initial surface by using singularity theory techniques ( cf .
thus we expect that principal curvatures may play an important role to study wave fronts from the differential geometric viewpoint .
in @xcite , murata and umehara showed that at least one principal curvature is unbounded near a singular point .
however , another principal curvature may be bounded @xmath0-function . in this paper , we give an explicit criterion for which a principal curvature becomes a bouded @xmath0-function near non - degenerate singular points of wave fronts in terms of geometric invariants ( theorem [ princi1 ] ) .
using this relation , we define analogies of ridge points for wave front ( definition [ ridge ] ) .
this kind of criteria for the case of cuspidal edges is given in ( * ? ? ?
* proposition 2.2 ) . as an application
, we consider singularities of parallel surfaces on wave fronts .
we studied parallel surfaces of cuspidal edges and gave criteria for swallowtails appearing on parallel surfaces of cuspidal edges in terms of geometric properties of cuspidal edges in @xcite .
however , we have not given criteria for other singularities which appear on parallel surfaces of cuspidal edges or wave fronts , in their differential geometric contexts .
thus we show relations between types of singularities of parallel surfaces on wave fronts and geometric properties of initial fronts ( theorem [ sing_para ] ) .
moreover , we give geometric relations between initial cuspidal edges and cuspidal edges which appear as singularities of parallel surfaces ( propositions [ para_kappa ] , [ negative ] and [ exactcusp ] ) .
in addition , we consider constant principal curvature ( cpc ) lines near singular points of wave fronts .
using parallel surfaces , we define special points ( landmark in the sense of @xcite ) on cuspidal edge as cusps of cpc lines , which seems not to have appeared in the literature ( subsection [ cpc ] ) . finally , we treat the extended distance squared function on wave fronts . for the case of generic regular surfaces , singularities of extended distance
squared functions correspond to types of singularities of parallel surfaces ( cf .
* theorem 3.4 ) ) .
however , for wave fronts , the same statement does not hold , in fact , different kinds of singularities ( @xmath3-type ) will appear ( theorem [ d4front ] ) .
all maps and functions considered here are of class @xmath0 unless otherwise stated .
we recall some properties of wave fronts . for details , see @xcite .
a map @xmath4 is called a _ wave front _ ( or a _ front _ ) if there exists a unit normal vector @xmath5 to @xmath6 such that the pair @xmath7 gives an immersion , where @xmath8 is a domain and @xmath9 denotes the unit sphere in @xmath2 ( cf .
a map @xmath4 is called a _ frontal _ if just a unit normal vector @xmath5 to @xmath6 exists .
a point @xmath10 is said to be a _
singular point _ of @xmath6 if @xmath6 is not an immersion at @xmath10 .
we denote by @xmath11 the set of singular points of @xmath6 . for a frontal @xmath6 , the function @xmath12 as @xmath13
is called the _ signed area density function _ ( cf .
@xcite ) . by the definition of @xmath14
, @xmath15 holds .
we call @xmath16 a _ non - degenerate _ if @xmath17 .
let @xmath10 be non - degenerate .
then there exists a regular curve @xmath18 with @xmath19 such that @xmath20 locally parametrizes @xmath11 .
moreover , there exists a vector field @xmath21 such that @xmath22 along @xmath20 .
we call @xmath20 and @xmath21 the _ singular curve _ and the _ null vector field _ , respectively . a non - degenerate singular point @xmath10
is said to be of the _ first kind _ if @xmath23 is transverse to @xmath24 , that is , @xmath25 .
otherwise , it is said to be of the _ second kind _
( @xcite ) .
moreover , we call a non - degenerate singular point of the second kind _ admissible _ if the singular curve consists of points of the first kind except at @xmath10 . otherwise , we call @xmath10 the _ non - admissible_. let @xmath26 be a map - germ around @xmath10 . then @xmath6 at @xmath10 is a _ cuspidal edge _ if the map - germ @xmath6 is @xmath27-equavalent to the map - germ @xmath28 at @xmath29 , @xmath6 at @xmath10 is a _ swallowtail _
if the map - germ @xmath6 is @xmath27-equivalent to the map - germ @xmath30 at @xmath29 , @xmath6 at @xmath10 is a _
cuspidal butterfly _
if the map - germ @xmath6 is @xmath27-equivalent to the map - germ @xmath31 at @xmath29 , @xmath6 at @xmath10 is a _ cuspidal lips _ if the map - germ @xmath6 is @xmath27-equivalent to the map - germ @xmath32 at @xmath29 and @xmath6 at @xmath10 is a _
cuspidal beaks _ if the map - germ @xmath6 is @xmath27-equivalent to the map - germ @xmath33 at @xmath29 , where two map - germs @xmath34 are _
@xmath27-equivalent _ if there exist diffeomorphism - germs @xmath35 on the source and @xmath36 on the target such that @xmath37 holds .
we note that generic singularities of fronts are cuspidal edges and swallowtails and generic singularities of one - parameter bifurcation of fronts are cuspidal lips / beaks , cuspidal butterflies and @xmath38 singularities in addition to above two ( see @xcite ) . here _
@xmath38 singularities _ are map - germs @xmath27-equivalent to @xmath39 at @xmath29 .
[ crit_front ] let @xmath40 be a front germ , @xmath5 a unit normal to @xmath6 and @xmath10 a corank one singular point . 1 .
suppose that @xmath10 is a non - degenerate singular point .
* @xmath6 at @xmath10 is @xmath27-equivalent to a cuspidal edge if and only if @xmath41 .
* @xmath6 at @xmath10 is @xmath27-equivalent to a swallowtail if and only if @xmath42 and @xmath43 .
* @xmath6 at @xmath10 is @xmath27-equivalent to a cuspidal butterfly if and only if @xmath44 and @xmath45 .
2 . suppose that @xmath10 is a corank one degenerate singular point .
* @xmath6 at @xmath10 is @xmath27-equivalent to a cuspidal lips if and only if @xmath46 .
* @xmath6 at @xmath10 is @xmath27-equivalent to a cuspidal beaks if and only if @xmath43 and @xmath47 .
here @xmath14 is the signed area density function , @xmath21 the null vector field and @xmath48 the hessian matrix of @xmath14 .
cuspidal edges are non - degenerate singular points of the first kind . on the other hand , swallowtails and cuspidal butterflies
are of the admissible second kind ( cf .
thus generic singularities of fronts are admissible .
we note that there is a criterion for a _ cuspidal cross cap _ which appears on a frontal surface defined as a map - germ @xmath27-equivalent to @xmath49 at @xmath29 ( ( * ? ? ?
* theorem 1.4 ) ) .
we recall behavior of curvatures of fronts near non - degenerate singular points @xmath10 .
let @xmath4 be a front and @xmath5 a unit normal vector .
let @xmath50 and @xmath51 denote the gaussian and the mean curvature of a front @xmath6 .
it is known that @xmath51 is unbounded near @xmath10 ( ( * ? ? ?
* corollary 3.5 ) ) . on the other hand , for the gaussian curvature @xmath50
, it is known that @xmath50 is bounded near @xmath10 if and only if the second fundamental form vanishes along the singular curve @xmath20 ( ( * ? ? ?
* theorem 3.1 ) ) . for @xmath50 and @xmath51 ,
the notions called the _ rationally bounded _ and the _ rationally continuous _ are introduced in @xcite .
next we recall behavior of principal curvature maps of a front @xmath6 at singular points .
let us assume that there are no umbilic points on @xmath52 .
then there exists a local coordinate system @xmath53 centered at @xmath10 such that @xmath54 and @xmath55 @xmath56resp . @xmath57 and @xmath58 are linearly dependent on @xmath59 . in particular , the pair @xmath60 ( resp .
@xmath61 ) does not vanish at the same time ( ( * ? ? ?
* lemma 1.3 ) ) .
such a coordinate system is called a _ principal curvature line coordinate _ introduced in @xcite . for this local coordinate system @xmath53
, we define the maps @xmath62 @xmath63 which are called the _ principal curvature maps _ ( @xcite ) as the proportional ratio of the real projective line @xmath64 by @xmath65,\quad\lambda_2=[-\nu_v : f_v].\ ] ] [ bhv_pr_map ] let @xmath4 be a front and @xmath66 be the principal curvature maps . then @xmath67 is a singular point if and only if either @xmath68 $ ] or @xmath69 $ ] holds
. by fact [ bhv_pr_map ] , one principal curvature function of a wave front is bounded and the other is unbounded near a singular point .
let @xmath4 be a frontal , @xmath67 a non - degenerate singular point and @xmath5 a unit normal vector .
then we can take the following local coordinate system around @xmath10 .
[ adapt_1 ] a local coordinate system @xmath53 centered at a singular point of the first kind ( resp . of the second kind ) @xmath10 is called _ adapted _ if it is compatible with the orientation of @xmath52 and satisfies the following conditions : + ( 1 ) the @xmath70-axis is the singular curve , + ( 2 ) @xmath71 ( resp .
@xmath72 with @xmath73 ) gives the null vector field on the @xmath70-axis , + ( 3 ) there are no singular points other than the @xmath70-axis .
let @xmath10 be a cuspidal edge and @xmath53 an adapted coordinate system centered at @xmath10 . since @xmath74 along the @xmath70-axis , there exists a map @xmath75 such that @xmath76 .
we note that @xmath77 holds along the @xmath70-axis .
since @xmath78 on the @xmath70-axis by fact [ crit_front ] , the pair @xmath79 gives a frame ( cf .
@xcite ) .
[ wein1 ] it holds that @xmath80 where @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 and @xmath86 . for cuspidal edges ,
several geometric invariants are studied ( for example , see @xcite ) . by using an adapted coordinate system @xmath53 and the frame @xmath79
, we set the following invariants along the @xmath70-axis : @xmath87 @xmath88 , @xmath89 , @xmath90 and @xmath91 are called the _ singular curvature _ , the _ limiting normal curvature _ , the _ cuspidal curvature _ and the _ cusp - directional torsion _ , respectively .
see @xcite for details of their geometric meanings .
we note that these invariants can be defined on frontals with singular points of the first kind .
[ cuspidalcurv ] under the above settings , @xmath89 , @xmath90 and @xmath91 can be expressed as @xmath92 along the @xmath70-axis , where @xmath93 depends on the orientation of the frame @xmath79 .
one can check that @xmath89 can be expressed as above by defintions of functions .
we show @xmath90 and @xmath91 can be written as above formulas .
since @xmath5 is perpendicular to both @xmath54 and @xmath94 , @xmath5 can be written as @xmath95 .
first , we show that @xmath90 can be written as above .
we note that @xmath96 holds on the @xmath70-axis .
since @xmath97 , @xmath90 on the @xmath70-axis is expressed as @xmath98 on the @xmath70-axis .
next , we consider @xmath91 . since @xmath99 and @xmath100 on the @xmath70-axis
, we see that @xmath101 it is known that @xmath102 does not vanish if @xmath10 is a cuspidal edge ( cf .
* lemma 2.11 ) ) .
in particular , @xmath103 never vanishes on the @xmath70-axis by lemma [ cuspidalcurv ] .
take an adapted coordinate system @xmath53 with @xmath104 .
then @xmath105 holds on the @xmath70-axis ( see lemma [ cuspidalcurv ] ) . if @xmath106 , @xmath107 holds .
we define the following functions on @xmath108 as @xmath109 where @xmath110 these functions are well - defined on @xmath108 .
we remark that @xmath111 ( resp .
@xmath112 ) becomes @xmath113 ( resp .
@xmath114 ) if we change @xmath5 to @xmath115 .
let @xmath50 and @xmath51 be the gaussian and the mean curvature of @xmath6 defined on @xmath108 . then @xmath116 and @xmath117 hold .
thus we may treat @xmath111 and @xmath112 as _ principal curvatures _ of @xmath6 defined on @xmath108 . here
@xmath50 and @xmath51 can be expressed as @xmath118 on the set of regular points .
we note that @xmath119 hold on the set of regular points .
let @xmath4 be a frontal , @xmath10 a non - degenerate singular point of the second kind and @xmath5 a unit normal vector to @xmath6 .
we fix an adapted coordinate system @xmath53 in the following ( see definition [ adapt_1 ] ) . taking a null vector field @xmath21
, there exists a function @xmath120 on the @xmath70-axis with @xmath73 so that @xmath72 ( see @xcite ) .
( we note that @xmath121 holds on the @xmath70-axis , namely , @xmath122 if @xmath10 is of the non - admissible . )
thus it follows that @xmath123 holds along the @xmath70-axis . on the other hand , since the @xmath70-axis gives the singular curve , there exists a @xmath0-function @xmath75 such that @xmath124 .
hence we have @xmath125 .
we remark that @xmath126 and @xmath5 are linearly independent since @xmath127 holds on the @xmath70-axis .
[ weinsec ] under the adapted coordinate system @xmath53 , @xmath55 and @xmath128 on @xmath59 can be written as @xmath129 where @xmath130 , @xmath131 , @xmath132 , @xmath133 , @xmath134 and @xmath135 .
we now define two @xmath0-functions on @xmath108 by @xmath136 where @xmath137 since the gaussian curvature @xmath50 and the mean curvature @xmath51 of @xmath6 satisfy @xmath116 and @xmath138 , we may regard @xmath139 as _ principal curvatures _ of @xmath6 on @xmath108 , where @xmath50 and @xmath51 are written as @xmath140 on @xmath108 .
we remark that @xmath119 hold on the set of regular points .
we put @xmath141 .
this is a @xmath0-function on @xmath59 .
it follows that @xmath142 holds along the @xmath70-axis ( cf .
we note that @xmath143 holds .
it is known that @xmath144 does not vanish on the @xmath70-axis if and only if @xmath6 is a front ( ( * ? ? ?
* proposition 3.2 ) ) .
we set @xmath145 this is a geometric invariant called the _ normalized cuspidal curvature _ defined in @xcite . by and
the definition of @xmath146 , we see that @xmath147 and @xmath148 hold if @xmath6 is a front .
[ limnormalsec ] under the above conditions , the limiting normal curvature @xmath89 can be written as @xmath149 at @xmath10 if @xmath10 is of the admissible second kind . by (
* proposition 1.9 ) , @xmath150 , @xmath151 and @xmath152 , we get the conclusion .
in this subsection , we consider boundedness of principal curvatures of fronts by using the above arguments . [ princi1 ]
let @xmath4 be a front and @xmath10 a non - degenerate singular point .
* let @xmath10 be a cuspidal edge .
* suppose that @xmath153 . if @xmath154 , then the principal curvature @xmath111 is a bounded @xmath0-function at @xmath10 .
moreover , @xmath155 .
* suppose that @xmath156 . if @xmath157 , then the principal curvature @xmath111 is a bounded @xmath0-function at @xmath10 .
moreover , @xmath155 @xmath56resp .
* let @xmath10 be of the second kind . if @xmath159 , then the principal curvature @xmath111 is a bounded @xmath0-function at @xmath10 .
moreover , @xmath155 if @xmath10 is an admissible .
converses are also true . moreover ,
if one of @xmath139 is bounded at @xmath10 , then the another is unbounded .
we prove the first asserion .
let @xmath4 be a front and @xmath10 a cuspidal edge .
take an adapted coordinate system @xmath53 centered at @xmath10 we show the case of @xmath104 . in this case , @xmath105 holds along the @xmath70-axis . for the case of @xmath106
, one can show similarly .
we now assume that @xmath154
. then @xmath160 by .
since @xmath161 and , we see that @xmath111 is a bounded @xmath0-function on @xmath59 and @xmath162 holds at @xmath10 .
conversely , we assume that the principal curvature @xmath111 is a bounded @xmath0-function near @xmath10 . in this case
, it follows that @xmath163 is positive along the @xmath70-axis .
this implies that @xmath90 is positive along the @xmath70-axis by .
unboundedness of @xmath112 near @xmath10 follows from the fact that the mean curvature is unbounded near @xmath10 .
next , we prove the second assertion .
take an adapted coordinate system @xmath53 centered at a non - degenerate singular point of the second kind @xmath10 .
suppose that @xmath164 .
it follows that @xmath165 holds near @xmath10 from . since @xmath166 , @xmath167 and @xmath165 along the @xmath70-axis
, it follows that @xmath168 and @xmath169 hold on the @xmath70-axis .
hence by , we have @xmath170 along the @xmath70-axis , and @xmath111 is a @xmath0-function . by lemma [ limnormalsec ] , we see that @xmath171 at @xmath10 if @xmath10 is an admissible . the converse and unboundedness
can be shown by using similar arguments to ( 1 ) . by theorem [ princi1 ] , one of @xmath139 of fronts
can be defined as a bounded @xmath0-function near non - degenerate singular points .
this implies there is a principal vector with respect to such a principal curvature at the singular point .
hence we consider explicit representation of the principal vector under an adapted coordinate system .
let @xmath4 be a front , @xmath10 a singular point of the second kind and @xmath5 a unit normal vector to @xmath6 .
then we take an adapted coordinate system @xmath53 around @xmath10 .
assume that @xmath159 , namely , @xmath111 is a bounded @xmath0-function near @xmath10 in the following .
we investigate the principal vector relative to @xmath111 .
let @xmath172 and @xmath173 denote the first and the second fundamental matrices given by @xmath174 the principal vector @xmath175 with respect to @xmath111 is a never vanishing vector satisfying @xmath176 .
we can write this equation as @xmath177 we note that @xmath178 does not vanish at @xmath10 .
thus we can take the principal vector @xmath179 as @xmath180 by factoring out @xmath181 from .
for the case of cuspidal edges , the principal vector @xmath179 with respect to @xmath111 is given as follows @xcite : @xmath182 we can extend the notion of a line of curvature as follows .
the singular curve @xmath20 is a _ line of curvature _ if the principal vector @xmath179 is tangent to @xmath20 .
[ curvline ] let @xmath4 be a front , @xmath10 a non - degenerate singular point and @xmath20 the singular curve passing through @xmath10 .
then the following assertions hold@xmath183 1 .
suppose that @xmath10 is a cuspidal edge .
then @xmath20 is a line of curvature of @xmath6 if and only if @xmath91 vanishes identically along @xmath20 .
2 . suppose that @xmath10 is of the second kind .
then @xmath20 can not be a line of curvature .
first , we show assertion ( 1 ) .
take an adapted coordinate system @xmath53 centered at a cuspidal edge @xmath10 satisfying @xmath104 .
assume that @xmath111 is bounded on @xmath59 .
then the principal vector @xmath175 relative to @xmath111 is given by .
since @xmath184 on the @xmath70-axis , @xmath185 can be written as @xmath186 along the @xmath70-axis by lemma [ cuspidalcurv ] .
thus @xmath185 vanishes on the @xmath70-axis if and only if @xmath91 vanishes along the @xmath70-axis , and we get the conclusion .
next , we show ( 2 ) .
take an adapted coordinate system @xmath53 around @xmath10 and assume that @xmath159 holds . in this case
, @xmath111 is bounded on @xmath59 and the principal vector @xmath175 of @xmath111 is given as . the second component @xmath185
is written as @xmath187 along the @xmath70-axis .
thus we have @xmath188 at @xmath10 .
this implies that the @xmath70-axis can not be the line of curvature . using the principal curvature @xmath111 and the principal vector @xmath179 relative to @xmath111 ,
we define ridge points for @xmath6 .
ridge points play important role to study parallel surfaces .
[ ridge ] under the above settings , a point @xmath10 is called a _ ridge point _
if @xmath189 holds , where @xmath190 denotes the directional derivative of @xmath111 with respect to @xmath179 .
moreover , a point @xmath10 is called a _
k - th order ridge point _ if @xmath191 and @xmath192 hold , where @xmath193 means the @xmath194-th directional derivative of @xmath111 with respect to @xmath179 .
ridge points for regular surfaces were first studied deeply by porteous @xcite .
he showed that ridge points correspond to @xmath195 singular points , that is , cuspidal edges of caustics . for more details on ridge points ,
see @xcite .
for the case of regular surfaces , principal curvatures relate singularities of parallel surfaces . in this section , we consider singularities of parallel surfaces of fronts and give criteria in terms of principal curvatures and other geometric properties .
swallowtails on parallel surfaces of cuspidal edges are studied in @xcite . here
we give criteria for other singularities on parallel surfaces of fronts . in this subsection
, we shall deal with fronts which have singular points of the second kind ( swallowtails , for example ) .
needless to say , the following arguments can be applied to the case of cuspidal edges .
let @xmath4 be a front , @xmath5 a unit normal to @xmath6 and @xmath67 a non - degenerate singular point of the second kind .
then the _ paralle surface @xmath196 of @xmath6 _ is defined by @xmath197 , where @xmath198 is constant .
we note that @xmath196 is also a front since @xmath5 is a unit normal to @xmath196 .
[ singsetpara ] let @xmath4 be a front , @xmath5 its unit normal vector and @xmath10 a non - degenerate singular point of @xmath6 .
suppose that @xmath111 is a bounded @xmath0-function near @xmath10 and @xmath199 .
then @xmath10 is a singular point of @xmath196 if and only if @xmath200 .
moreover , @xmath10 is non - degenerate if and only if @xmath10 is not a critical point of @xmath111 .
we show the case that @xmath10 is of the second kind .
let @xmath53 be an adapted coordinate system centered at @xmath10 with the null vector field @xmath72 .
then the signed area density function @xmath201 of @xmath196 can be written as @xmath202 by lemma [ weinsec ] , where @xmath203 .
since @xmath204 does not vanish at @xmath10 , @xmath10 is a singular point of @xmath196 if and only if @xmath200 holds .
thus we may treat @xmath205 as the signed area density function of @xmath196 .
non - degeneracy follows @xmath206 .
[ sing_para ] let @xmath207 be a front and @xmath10 be a non - degenerate singular point .
suppose that the principal curvature @xmath111 is a bounded @xmath0-function on @xmath52 and @xmath199 .
then for the parallel surface @xmath196 with @xmath200 , the following conditions hold . 1 .
assume @xmath208 .
then the following hold@xmath183 * the map - germ @xmath196 at @xmath10 is @xmath27-equivalent to a cuspidal edge if and only if @xmath10 is not a ridge point of @xmath6 . *
the map - germ @xmath196 at @xmath10 is @xmath27-equivalent to a swallowtail if and only if @xmath10 is a first order ridge point of @xmath6 . *
the map - germ @xmath196 at @xmath10 is @xmath27-equivalent to a cuspidal butterfly if and only if @xmath10 is a second order ridge point of @xmath6 .
2 . assume @xmath209 .
then the following hold@xmath183 * the map - germ @xmath196 at @xmath10 is @xmath27-equivalent to a cuspidal lips if and only if @xmath210 and @xmath211 hold .
* the map - germ @xmath196 at @xmath10 is @xmath27-equivalent to a cuspidal beaks if and only if @xmath10 is a first order ridge point of @xmath6 , @xmath210 and @xmath212 hold .
+ here @xmath213 is the hessian matrix of @xmath111 at @xmath10 .
let @xmath4 be a front , @xmath67 a non - degenerate singular point of the second kind and @xmath5 a unit normal vector .
then we take an adapted coordinate system @xmath53 around @xmath10 . by lemma [ singsetpara ]
, we can take the signed area density function of parallel surface @xmath196 with @xmath200 as @xmath214 .
first , we prove the assertion ( 1 ) . in this case
, @xmath215 is a smooth curve near @xmath10 and there exists a null vector field @xmath216 of @xmath196 .
we set @xmath217 , where @xmath218 are functions on @xmath59 . by lemma
[ weinsec ] , @xmath219 on @xmath220 is equivalent to @xmath221 holds on @xmath222 .
thus the null vector field @xmath216 can be taken as the principal vector @xmath179 as in with respect to @xmath111 restricted to @xmath222 . under these conditions , the equation @xmath223 holds for some natural number @xmath224 .
thus we have the assertion ( 1 ) by fact [ crit_front ] ( 1 ) .
next , we prove ( 2 ) . in this case
, @xmath225 vanishes at @xmath10 .
we consider the rank of @xmath226 at @xmath10 .
the jacobian matrix @xmath227 of @xmath196 is @xmath228 at @xmath10 , where @xmath229 since @xmath230 , it follows that @xmath231 , when @xmath200 , and it implies that @xmath210 .
thus there exists a non - zero vector field @xmath216 near @xmath10 such that if @xmath232 then @xmath219 holds at @xmath233 .
we can take the principal vector @xmath179 with respect to @xmath111 as @xmath216 , then @xmath234 .
moreover , we see that @xmath235 , @xmath236 , @xmath237 .
thus we have @xmath238 . by using fact [ crit_front ] ( 2 ) and the definition of ridge points
, we have the conclusion .
this theorem implies that the behavior of a bounded principal curvature of fronts determines the types of singularities appearing on parallel surfaces . for regular surfaces and whitney umbrellas ,
similar results are obtained in @xcite . by in the proof of theorem [ sing_para ] and (
* theorem 1.1 ) , we see that a parallel surface @xmath196 does not have @xmath239 singularity at @xmath10 .
we shall consider geometric properties of parallel surfaces of cuspidal edges .
in particular , we deal with the case that parallel surfaces have cuspidal edges . for cuspidal edges ,
the following normal form is obtained by martins and saji @xcite .
[ normal_cusp ] let @xmath240 be a map - germ and @xmath29 a cuspidal edge .
then there exist a diffeomorphism - germ @xmath241 and an isometry - germ @xmath242 satisfying @xmath243 where @xmath244 and @xmath245 with @xmath246 smooth functions .
let @xmath4 be a normal form as in .
suppose that @xmath247 and @xmath248 , that is , @xmath111 is a bounded @xmath0-function near @xmath29 and @xmath249 .
[ para_kappa ] let @xmath4 be a normal form of a cuspidal edge as in and @xmath196 its parallel surface , where @xmath250 .
we assume that @xmath196 has a cuspidal edge at @xmath29 , that is , @xmath251 .
then it holds that the limiting normal curvature @xmath252 and the singular curvature @xmath253 can be written as @xmath254 at @xmath29 , where @xmath255 is determined as below .
since @xmath256 implies @xmath29 is a ridge point , if @xmath256 holds , then @xmath196 has a swallowtail or worse at @xmath29 ( cf .
@xcite ) .
let @xmath257 be a singular curve of the parallel surface @xmath196 of @xmath6 satisfying @xmath258 and @xmath259 where @xmath200 and @xmath260 is a sufficiently small .
setting @xmath261 , we see @xmath262 since @xmath263 ( see lemma [ wein1 ] ) .
thus we have that @xmath264 .
next , we consider the singular curvature @xmath253 .
we now assume @xmath265
. then @xmath266 can be locally parametrized by the form @xmath267 around @xmath29 by the implicit function theorem . since @xmath268 , we have @xmath269 and @xmath270 by differentiating @xmath271 .
thus @xmath272 hold at @xmath29 ( see and ( * ? ? ?
* lemma 2.2 ) ) .
we set @xmath273 ( resp .
@xmath274 ) if @xmath190 and @xmath275 have the same sign ( resp .
opposite sign ) .
thus we have @xmath276 at @xmath29 .
if @xmath277 at @xmath29 , we have the same formulation as above .
let @xmath4 be a front , @xmath5 a unit normal vector and @xmath10 as non - degenerate singular point .
suppose that @xmath111 is bounded at @xmath10 and @xmath199 .
we set @xmath214 .
the zero - set of this function gives the singular curve of the parallel surface @xmath196 of @xmath6 , where @xmath200 ( lemma [ singsetpara ] ) .
we call the curve given by @xmath278 a _ constant principal curvature @xmath56cpc@xmath279 line with the value of @xmath280 _ ( cf .
@xcite ) . in this case
, the cpc line is a regular curve since @xmath281 . in @xcite ,
cpc lines for regular surfaces and whitney umbrellas , and relations between singularities of parallel surfaces and the behavior of cpc lines are investigated . for intrinsic properties of whitney umbrellas ,
see @xcite .
let @xmath4 be a front , @xmath67 a cuspidal edge and assume that @xmath111 is bounded near @xmath10 .
the condition @xmath282 at @xmath10 implies that the cpc line is tangent to the null vector field @xmath21 of @xmath6 at @xmath10 .
moreover , the image @xmath283 of the set of singular points of the parallel surface @xmath196 by @xmath6 is cusped at @xmath10 .
we call such a point an _ exactly cusped point for the constant principal curvature @xmath56cpc@xmath279 line_. [ negative ] let @xmath4 be a front and @xmath10 a cuspidal edge .
suppose that @xmath111 @xmath56resp .
@xmath284 is bounded at @xmath10 . then @xmath285 @xmath56resp .
@xmath286 implies @xmath287 . by using a normal form as in
, we have @xmath288 at @xmath29 . since the null vector is @xmath71 for a normal form , the relation @xmath289 holds .
hence @xmath290 if and only if @xmath291 .
this implies that @xmath292 thus we obtain the assertion .
relations between the gaussian curvature and the singular curvature are stated in ( * ? ? ?
* theorem 3.1 ) .
[ exactcusp ]
let @xmath4 be a front , @xmath10 a cuspidal edge , @xmath20 a singular curve and @xmath21 a null vector field .
assume that @xmath111 is bounded near @xmath10 , @xmath199 and @xmath10 is not a ridge point of @xmath6 .
then the cusp - directional torsion @xmath293 of @xmath196 vanishes at @xmath10 if and only if @xmath294 vanishes at @xmath10 , namely , @xmath10 is an exactly cusped point , where @xmath200 .
let @xmath4 be a normal form as in and @xmath266 be a singular curve of @xmath196 satisfting @xmath295 .
we assume that @xmath265
. then we can take @xmath267 .
let @xmath296 denote a vector field tangent to @xmath266 , where @xmath269 .
the pair @xmath297 gives an adapted pair of vector fields in the sense of @xcite .
moreover , @xmath298 holds at @xmath29 . by (
* ( 5.1 ) ) , we have @xmath299 comparing with and , we obtain the result .
we now consider the case that @xmath300 at @xmath10 . since this is equivalent to @xmath301 at @xmath10 , we call such a point an _ extrema of the limiting normal curvature _ @xmath89 .
therefore we have three special points ( landmarks in the sence of @xcite ) on cuspidal edges which have special relations between the singular curve and the cpc line ( see figure [ fig : cpc ] ) .
it seems that exactly cusped points have not appeared in the literature .
c .,title="fig:",width=132 ] exactly cusped point .,title="fig:",width=132 ] extrema of @xmath89 .,title="fig:",width=132 ] ridge point
we consider the extended distance squared functions on fronts .
we study relations between singularities of extended distance squared functions and principal curvatures . for singularities of distance squared functions on surfaces with other corank one singularities , see @xcite .
let @xmath4 be a front , @xmath5 a unit normal vector and @xmath10 be a non - degenerate singular point of the second kind .
( for cuspidal edges , see @xcite . )
assume that @xmath111 is bounded at @xmath10 and @xmath199 in this section .
let us take an adapted coordinate system @xmath53 centered at @xmath10 with the null vector field @xmath72 . in this case
, @xmath310 follows from . by direct computations , we have @xmath311 and @xmath312 .
since @xmath313 at @xmath10 , we get the conclusion .
take @xmath309 .
we are interested in the case of @xmath314 , because @xmath307 corresponds to the image of a singular point of a parallel surface @xmath196 with @xmath200 , that is , @xmath307 coincides with a _ focal point _ of @xmath6 at @xmath10 .
in such a case , @xmath306 measures contact of @xmath6 with the _ principal curvature sphere _ at @xmath10 ( cf .
@xcite ) .
take an adapted coordinate system @xmath53 around @xmath10 . by lemma [ singpsi ]
, we see that @xmath319 at @xmath10 . by direct calculations
show @xmath320 thus we have @xmath321 at @xmath10 since @xmath322 .
moreover , it follows that @xmath323 at @xmath10 since @xmath324 .
thus we have the assertion . proposition [ rankzero ] implies that @xmath306 may have a @xmath239 singularity at @xmath10 if @xmath307 coincides with the focal point of @xmath6 at @xmath10 , where a function - germ @xmath325 has a _
@xmath239 singularity _ at @xmath29
if @xmath326 is @xmath327-equivalent to the germ @xmath328 at @xmath29 ( cf .
* and 265 ) ) .
therefore we consider the condition that @xmath306 has a @xmath239 singularity at @xmath10 in terms of geometric properties of @xmath6 . for cuspidal edges ,
suppose that @xmath111 is bounded near @xmath10 and @xmath199 .
then the function @xmath306 with @xmath315 and @xmath314 has a @xmath239 singularity if and only if @xmath329 ( ( * ? ? ?
* theorem 3.3 ) ) . here
@xmath330 is a geometric invariant called the _ edge inflectional curvature _ defined in ( * ? ? ?
* section 5.3 ) .
we remark that oset sinha and tari @xcite study singularities of height functions and orthogonal projections of cuspidal edges .
let @xmath4 be a front and @xmath10 a sngular point of the second kind .
for a function @xmath302 , set @xmath331 it is known that the function @xmath306 is @xmath327-equivalent to @xmath332 ( resp .
@xmath333 ) if and only if @xmath316 and @xmath334 ( resp .
@xmath335 ) hold ( see ( * ? ? ?
* lemma 3.1 ) , see also ( * ? ? ?
* theorem 4.2 ) ) .
[ d4front ] let @xmath4 be a front and @xmath10 a singular point of the second kind .
suppose that @xmath111 is bounded near @xmath10 and @xmath199 .
then @xmath306 as in with @xmath315 and @xmath314 has a @xmath239 singularity at @xmath10 if and only if @xmath10 is not a ridge point of @xmath6 . to prove this theorem
, we take a special adapted coordinate system @xmath53 centered at @xmath10 called a _ strongly adapted coordinate system _ which satisfies @xmath336 at @xmath10 ( see ( * ? ? ?
* definition 3.6 ) ) . under this coordinate system
, we see that @xmath337 at @xmath10 since @xmath338 .
we prepare lemma .
we take a strongly adapted coordinate system @xmath53 around @xmath10 .
direct calculations show that @xmath341 hold at @xmath10 , where @xmath342 .
since @xmath343 , @xmath344 and @xmath345 at @xmath10 , it follows that @xmath346 and @xmath347 hold at @xmath10 .
thus @xmath348 as in can be written as @xmath349 this implies that @xmath339 if and only if @xmath350 .
we now deal with @xmath354 .
it follows that @xmath355 and @xmath356 on @xmath59 .
so @xmath357 holds . by lemma [ weinsec ]
, @xmath358 is written as @xmath359 at @xmath10 . on the other hand , @xmath360 at @xmath10 by lemma [ weinsec ] .
hence we have @xmath361 and @xmath362 at @xmath10 , where we used @xmath363 . by , and , @xmath364 can be written as @xmath365 at @xmath10 .
thus we have the assertion .
let us take a strongly adapted coordinate system @xmath53 centered at @xmath10 .
then we note that @xmath337 holds at @xmath10 .
the differentials @xmath275 and @xmath366 are given by @xmath367 at @xmath10 .
since the principal vector @xmath179 as in is written as @xmath368 at @xmath10 , we have @xmath369 comparing and in lemma [ del ] , @xmath370 , namely , @xmath10 is not a ridge point of @xmath6 if and only if @xmath350 .
this implies that the number @xmath348 defined as does not vanish by lemma [ del ] .
we remark that the condition that @xmath6 is a front in theorem [ d4front ] is needed for @xmath306 to have a @xmath239 singularity at @xmath10 .
in fact , for a frontal @xmath4 with a singular point of the admissible second kind @xmath10 , we have the following . [ d5frontal ] let @xmath4 be a frontal but not a front and @xmath10 a singular point of the admissible second kind .
then @xmath306 with @xmath371 and @xmath372 does not have a @xmath239 singularity at @xmath10 .
let us take an adapted coordinate system @xmath53 centered at @xmath10 with the null vector field @xmath72 . by the proof of lemma [ del ] , @xmath373 vanishes automatically if @xmath6 is not a front at @xmath10 .
99 , vol.1 , second edition , , ( 1999 ) , 243-268 . , ( 2008 ) , 827848 . ,
( 2012 ) , 387408 . , ( 2012 ) , 3567 .
4 ( 2015 ) , 34pp . , ( 2014 ) , 769785 . , , ( 2010 ) , 92-127 . , ( 2010 ) , 789849 . ,
( 2005 ) , 303351 . , ( 2007 ) ,
265299 . , ( 2015 ) , 105124 . .
( 2016 ) , 445462 .
, , ( 2005 ) , 303351 . , .
r. oset sinha and f. tari , _ on the flat geometry of the cuspidal edge _ , preprint , arxiv:1610.08702 . ,
( 1971 ) , 543564 .
second edision , , ( 2011),137147 . , ( 2008 ) , 259-280 . , ( 2009 ) , 731746 . , ( 2009 ) , 491529 . k. saji , m. umehara and k. yamada , _ the duality between singular points and inflection points on wave fronts _ , osaka math . j. * 47 * ( 2010 ) , 591607 . , ( 2012 ) , 285299 . , ( 2016 ) , 5262 . | we give criteria for which a principal curvature becomes a bounded @xmath0-function at non - degenerate singular points of wave fronts by using geometric invariants . as applications , we study singularities of parallel surfaces and extended distance squared functions of wave fronts .
moreover , we relate these singularities to some geometric invariants of fronts . |
the cosmic infrared background ( cib ) is the collective radiation emitted throughout cosmic history , including from sources inaccessible to current telescopic studies ( see review by kashlinsky , 2005 ) .
the latter category includes the very first luminous objects , currently a subject of intense investigations and great importance for astronomy and cosmology .
cib fluctuations can be more readily discerned than the actual mean level of the background allowing to overcome the significant galactic and solar system foregrounds at these wavelengths ( kashlinsky et al .
1996a , b , kashlinsky & odenwald 2000 ) .
it is generally believed now as a result of detailed numerical simulations of the formation of first structures in the standard cosmological model that the first objects to form in the universe , the so - called population iii stars , were very massive stars that occupied a brief era at epochs inaccessible to direct telescopic observations ( see review by bromm & et al .
because distant galaxies and pre - galactic structures are clustered , the cib has angular fluctuations with a distinct spectral and spatial signal including a possibly measurable contribution from the first stars era ( bond et al .
1986 , kashlinsky et al .
2004 , cooray et al .
the distribution on the sky of the luminous objects to form at early times should be considerably different from the cosmic pattern seen today , with the differences diverging toward large angular scales and being particularly prominent between 5@xmath5 to 1@xmath6 .
although the individual sources at very high @xmath7 are too faint to observe on their own , fluctuations in the intensity of the cosmic infrared background ( cib ) will reflect the distribution of those early objects after foreground sources are removed to sufficiently faint levels . in the course of the prior work we discovered significant source - subtracted cib fluctuations on scales as large as @xmath8 in deep _ spitzer _
irac ( 3.68 @xmath9 m ) data ( kashlinsky et al . 2005 , 2007a , b , c - hereafter kamm1,2,3,4 ) .
extensive details of our analysis and a multitude of tests it was subjected to along with a summary of requirements any reliable cib data analysis should meet in the presence of significant foreground and instrumental effects is given in arendt et al .
( 2010 , hereafter - akmm ) .
as thoroughly documented in akmm , we have been very careful with the processing of the individual irac exposures into complete deep mosaicked images .
akmm include details with the numerous thorough checks to confirm that we are not being misled by instrumental artifacts or variations of the zodiacal light .
we test that we see similar background fluctuations in various fields located at different ecliptic and galactic latitudes . while all our background fluctuation measurements have been derived from irac data , the recent results of matsumoto et al .
( 2011 ) find similar background fluctuations using completely independent data from the _ akari _ satellite , and with performing an independent analysis .
this is the further confirmation that the fluctuations are not caused by any hidden error in the irac data or our analysis .
the residual cib fluctuations remain after removing galaxies down to very faint levels and must arise from populations with a significant clustering component , but only low levels of the shot noise .
as suggested by kamm1 these cib fluctuations may originate in early populations .
further , it was demonstrated by kamm4 that there are no correlations between the source - subtracted irac maps and the faintest resolved sources observed with the _ hst _ acs at optical wavelengths , which likely points to the high-@xmath7 origin of the fluctuations , or at least to a very faint population not yet observed by other means .
the high-@xmath7 interpretation of the detected cib anisotropies has received strong further confirmation in the recent _ akari _ data analysis which , in addition , measured source - subtracted cib fluctuations at 2.4 and pointed out that the colors of the fluctuations require them being produced by highly redshifted very luminous sources ( matsumoto et al .
it is inherently important to advance the understanding of the nature and the epochs of the populations producing these cib fluctuations with new studies which can be obtained measuring the spatial distribution of the fluctuations at much larger angular scales .
this was one of the motivations of the seds program ( fazio et al .
2008 ) and the first results from it are presented below .
this study followed the procedures outlined in akmm and our previous publications ( kamm1 - 4 ) and the paper is organized as follows : sec .
[ sec : data ] describes the data and its assembly , followed by sec .
[ sec : fourier ] which lays out the quantities computed from the assembled data . in sec .
[ sec : results ] we present the results on the spatial spectrum of the source - subtracted fluctuation in the diffuse light remaining after subtracting the instrument noise .
[ sec : foregrounds ] presents comprehensive tests done to isolate the various non - cosmological contributions to the results and both low- and high-@xmath7 contributors are discussed in sec .
[ sec : interpretation ] , where it is shown that the signal can not be accounted for by the observed populations of known galaxies and requires either unknown new populations at low-@xmath7 made of low - luminosity systems and distributed very differently from the rest of the present day galaxies , or is dominated by the high-@xmath7 populations at epochs associated with first stars " and clustered according to the established concordance @xmath4cdm model .
the _ spitzer space telescope _ is a 0.85 m diameter telescope launched into an earth - trailing solar orbit in 2003 ( werner et al .
2004 , gehrz et al .
2007 ) . for nearly 6 years ,
as it was cooled by liquid he , its three scientific instruments provided imaging and spectroscopy at wavelengths from 3.6 to 160 @xmath9 m . in the time since the he supply was exhausted , _ spitzer _ has continued to provide 3.6 and 4.5 @xmath9 m imaging with its infrared array camera ( irac ) .
the irac camera has a @xmath10 field of view , and a pixel scale of @xmath11 , which slightly undersampled the instrument beam size of @xmath12 ( fwhm ) ( fazio et al .
2004a ) .
the seds program is designed to provide deep imaging at 3.6 and 4.5 over a total area of about 1 square degree , distributed over 5 well - studied regions ( fazio et al .
the area covered is about ten times greater than previous _
coverage at comparable depth .
while the main use of the seds data sets will be the investigation of the individually detectable and countable galaxies , the remaining backgrounds in these data are well - suited for cib studies , by virtue of their angular scale , sensitivity and observing strategies .
the first 2 of the seds fields to be completed have been used in this study : the ultradeep survey field ( uds ; program i d = 61041 ) and the extended groth strip ( egs ; program i d = 61042 ) .
the locations , sizes , and depths of these fields are listed in table 1 .
the fields are located at moderate to high galactic latitudes to minimize the number of foreground stars and the brightness of the emission from interstellar medium ( cirrus ) .
these fields also lie at relatively high ecliptic latitudes , which helps minimize the brightness and temporal change in the zodiacal light from interplanetary dust .
the observations of each field are carried out at three different epochs , spaced 6 months apart . for cib fluctuations analysis in the assembled maps
we have selected regions of approximately uniform exposure in _ both _ channel 1 and 2 .
these covered a square region of approximately 21@xmath5 on the side in the uds field and a rectangular region of @xmath13 in the egs field .
the procedure for map assembly is described in our previous papers ( kamm1 - 4 ) with an extensive summary including all the tests given in ( akmm ) .
below we briefly revisit the adopted formalism and its advantages over the standard methods when it comes to removing instrumental artifacts without introducing spurious correlations .
this is critically important if one were to robustly measure fluctuations signals as faint as that expected from first stars era , @xmath14 nw / m@xmath15/sr at arcminute scales .
the individual 100 sec frame time exposures are processed by the standard irc calibration pipeline , and then additionally corrected for the `` column - pulldown effect '' which affects the detector output in columns that contain very bright stars .
the individual frames are then further processed and combined into mosaicked images ( with a @xmath11 pixel scale ) using the self - calibration procedure that we have previously employed for this work .
the 3.6 and 4.5 @xmath9 m images are processed independently . at each wavelength , the frames are also processed in several different groups to provide multiple images that can be used to assess random and systematic errors .
one set of three groups is obtained by separating the data by each of the three epochs .
comparison of results between these three sets can provide indications of systematic errors induced by variations in the zodiacal light over 6 month intervals .
the other set of two groups is obtained by separating the full sequence of frames into the alternating even and odd frame numbers .
comparison of results from these `` a '' and `` b '' subsets provides a good diagnostic of the random instrument noise , because the a and b subsets only differ by a mean interval of @xmath16s .
we applied the least - squares self - calibration procedure described by ( fixsen et al .
the approach formalizes the calibration procedure by describing the data with parameters that include both the detector characteristics and the true sky intensity .
the derivation of these parameters via a least - squares algorithm yields an optimal solution for the calibration and the sky intensity . in this case
our chosen model is given by @xmath17 where @xmath18 represents the raw data from a single pixel of a single frame , @xmath19 is the sky intensity at location @xmath20 , @xmath21 is the offset for detector pixel @xmath22 , and @xmath23 is a variable offset for each of the 4 readouts ( alternate vertical columns of the detector ) and each frame .
this assumes that the sky intensity ( @xmath19 ) and the detector offsets ( @xmath21 ) are invariant during the course of the observations . for a data set with fixed frame times ( as our irac data ) , the detector dark current
is included in the @xmath21 term as it is indistinguishable from an offset . for data sets with multiple frame times ,
a relatively simple extension of this data model could be applied .
the variable offset @xmath23 can absorb time
dependent behavior of the detector , but only to the extent that it can be characterized with a single value per frame , or in some cases , a single value per readout per frame .
while technically the cib is the sum of all extragalactic emission , here we wish to exclude individually detectable objects and analyze the remaining cib which is produced by fainter sources .
the mosaicked images were cleaned of resolved sources in two stages . in the first stage ,
we construct and subtract an iterative `` model '' of all the sources in the image .
this is done using a variant of the clean algorithm , in which our iterative loop consists of ( a ) identifying the brightest pixel in the image , and ( b ) subtracting a scaled version of the instrument psf ( including the broad low - level wings ) to reduce this intensity by a set fraction ( chosen as 50% ) . subtracting only a fraction of a source at each iteration
allows the model to compensate for sources that are intrinsically extended in size , and for inaccuracies in the psf .
various criteria for selecting the final iteration are given in akmm ; all lead to very similar results . in order to compare the signal from different sky locations ,
the final iteration can be chosen to correspond to a given shot - noise level .
the second stage involves the construction of masks which are used to set areas at the locations of bright sources to zero .
if the model worked perfectly , this would not be necessary . however , small differences between the model s ideal psf and the effective psf of the mosaicked images lead to relatively large amplitude residuals at the locations of bright stars .
therefore we construct a mask which is used to zero all pixels above an effective @xmath24 surface brightness .
all 8 neighbors of any such pixel are also set to zero .
this clipping is done independently of the model - subtraction and the procedure is iterated until no new pixels above the fixed threshold remain .
the fraction of unmasked pixels is @xmath25 which enables robust fft analysis .
= 1 the histogram distributions of the remaining pixel fluctuations after removal of the modeled sources and masking are shown in fig .
[ fig : histo ] .
they are to a good accuracy gaussian with very small residual skewness , @xmath26 , parameters ( @xmath27 ) . in order to evaluate the random noise level of the maps ,
alternate calibrated frames for each dataset were mapped into separate `` a '' and `` b '' mosaics .
the difference of the a and b mosaics should eliminate true celestial sources and stable instrumental effects and reflect only the random noise of the observations .
the power spectrum of the instrument noise is then subtracted from the power spectrum of the assembled maps to give the measured signal of the remaining source - subtracted cib fluctuations in the maps .
the maps under study are clipped and masked of the resolved sources , yielding the fluctuation field , @xmath28 .
its fourier transform , @xmath29 is calculated using the fft .
the power spectrum is @xmath30 , with the average taken over all the independent fourier elements corresponding to the given @xmath31 .
a typical rms flux fluctuation is @xmath32 on the angular scale of wavelength @xmath33 .
the correlation function , @xmath34 , is uniquely related to @xmath35 via fourier transformation .
if the fraction of masked pixels in the maps is too high ( e.g. @xmath36 for irac maps as detailed in kamm1 - 4/akmm ) , the large - scale map properties can not be computed using the fourier transform and instead the maps must be analyzed by direct calculation of @xmath37 , which is immune to mask effects ( kashlinsky 2007 ) .
several quantities are used in computations below .
the power spectrum in a single band @xmath38 is defined as @xmath39 .
it is computed after averaging all independent fourier elements which lie inside the radial interval @xmath40 $ ] .
since the flux is a real quantity , only one half of the fourier plane is independent , so that at each @xmath31 there are @xmath41 independent measurements of @xmath42 out of a full ring with @xmath43 data .
because we will compare two fields of different configurations , the fourier space of each was binned at the same grid of @xmath31 , the mean power was then evaluated and the relative ( poissonian ) errors on that determination were computed as @xmath44 . after demonstrating that different fields have statistically similar power spectrum ,
the overall power at each @xmath31 was averaged over the different fields with each field weighted by the relative statistical uncertainty defined by its configuration ( see below ) , i.e. @xmath45 .
this is equivalent to simply averaging over all @xmath46 available from each field @xmath47 .
we characterize the similarity ( or not ) of the fluctuations measured in different channels , or at different epochs of observations .
a quantity of further interest in this context is the cross - power , which is the fourier transform of the cross - correlation function @xmath48 .
the cross - power spectrum is then given by @xmath49 with @xmath50 standing for the real , imaginary parts .
note the cross - power of a real quantity , such as the flux fluctuation , is always real , but unlike the single ( auto- ) power spectrum the cross - power can be both positive and negative .
the presence ( or not ) of the same population at two channels , @xmath51 , can then be probed via _ coherence _ defined as @xmath52}$ ] . if @xmath53 then at the two channels the same populations produce the diffuse signal and vice versa .
the clipped and cleaned maps were fourier transformed and power spectra evaluated .
visual and numerical inspection of the maps did not show any significant presence of artifacts , stray light , muxbleed etc . in order to average over the different fields
the fourier space was binned at the same set of central wavenumbers , @xmath31 , in both configurations .
the values of @xmath31 chosen for the final @xmath54 evaluations are shown in the right panel of fig .
[ fig : histo ] which displays the number of independent fourier elements which is available for power spectrum evaluations . since the relative error on the determined @xmath54 is @xmath55^{-\half}$ ] ,
the power spectrum is not determined highly accurately at the outer 2 bins of the egs field ; however , when combined with the independent determination in the uds field , the accuracy of the averaged @xmath54 is acceptable ( @xmath56 ) at scales @xmath57 .
only the largest scale of @xmath58 probed with the egs field alone has @xmath59 and is not measured well .
the instrument noise was evaluated from the @xmath60 maps , where @xmath61 and @xmath62 correspond to the sequences of alternating odd / even aors .
= 0.8 figure [ fig : power ] shows the total power remaining in the source - subtracted maps and the noise . at small scales ( within the beam at @xmath63 )
the noise contributes significantly , as it should , but there is a clear excess in the fluctuation power at larger scales with the noise becoming progressively smaller with its contribution reducing to negligible at scales greater than @xmath64 .
the resulting power spectra of the residual cib after the above subtraction and masking of resolved sources is shown in fig .
[ fig : cib_fields ] . here
we have subtracted the instrumental noise power by deducting the power spectra of the differences of the a and b mosaics [ ( a - b ) ] , which are displayed in fig .
[ fig : power ] .
the instrumental noise power is comparable to the the cib power at scales @xmath65 , but is much smaller than the power at larger angular scales . at both 3.6 and 4.5 @xmath9 m ,
the power spectra are in good agreement for egs and uds fields , except perhaps at the largest angular scale ( @xmath66 ) where the small number of effective measurements leads to large uncertainties on the power in each individual field .
= 1 in order to check for the presence of the same source - subtracted cib at both 3.6 and 4.5 @xmath9 m , we have computed the cross - correlation power spectrum between the two channels
. the right panel of the figure shows the resultant cross - correlation power spectra .
their similarity to the full power spectra confirm that the fluctuations are correlated in wavelength for both fields at all angular scales .
similar amplitudes would not be expected if the signal was dominated by noise , or by instrumental artifacts that occur independently at each wavelength .
the correlation of many instrumental effects are mitigated by the fact that irac s 3.6 and 4.5 @xmath9 m channels use separate optical systems and detectors to simultaneously observe fields separated by @xmath67 .
the results in fig . [ fig : cib_fields ] are in excellent agreement with our earlier measurements in five additional fields ( kamm1 - 2 ) as shown in appendix a , but are now measured to significantly more accurately and extending to much greater angular scales .
the isotropy of the signal ( i.e. the consistency of results in different fields ) by itself suggests a cosmological origin of the fluctuations . nevertheless , we describe below a multitude of additional tests that we have performed in order to verify that non - cosmological sources can not explain the measured signal .
the data were analyzed separately after combining the aors in each of the 3 epochs of observations , e1-e3 .
epochs e1 and e2 ( and e2 and e3 ) are separated by @xmath68 months and the detector orientation changes by @xmath69 between each pair of epochs .
if the signal arises from the detector , rather than the sky , the fluctuations in each of the epochs should not correlate . additionally ,
if zodiacal light fluctuations contribute significantly to the measured signal , the measured fluctuations should differ appreciably between the epochs .
= 1 the random instrument noise is negligible at scales @xmath70 .
we cross - correlated the power spectra at epochs separated by 6 months when the detector orientation is rotated by 180@xmath6 .
instrument artifacts that remain in a fixed geometry with respect to the detector , such as stray light and ghost images , are thus projected on to the field in very different patterns at the different epochs .
the cross - correlations power spectra are shown in fig .
[ fig : cross ] and demonstrate that the signal comes from the sky rather than the detector .
one of the potential difficulties in measurements of backgrounds and their structure is stray light . stray
light will cause an apparent nonuniform increase in the background across the detector .
the intensity and the pattern of the stray light depends on the distribution of sources _ outside _ of the field of view being observed . in the case of the irac 3.6 and 4.5 @xmath9 m detectors , sources that lie @xmath71 from the top edge of the detector
can scatter light back onto the detector @xcite .
the geometry of the scattering leads to two concentrated regions of stray light in the upper left and right portions of the detector .
uniform diffuse illumination , such as the zodiacal light , leads to a fixed pattern on the detector .
an individual bright source leaves a more concentrated and irregular patch of stray light , but at a location that is well defined given the known source position .
an example of this is shown in the left and center panels of fig .
[ fig : stray ] .
thus individual irac exposures have been masked to exclude regions that are potentially affected by the stray light of stars identified by brightness and color thresholds from the 2mass catalog @xcite .
however , there remains some question of whether or not the stray light from sources at fainter levels might yield some structure in the observed background .
to address this question , we created new versions of the 3.6 @xmath9 m mosaic of the uds field from two subsets for the data .
the first version is made only with data from the upper half of detector , which can be affected by stray light if the data are insufficiently masked .
the second version is made from only from data in the lower half of the detector which is largely unperturbed by stray light .
the difference between these two mosaics would reveal if there were systematic differences between the upper and lower halves of the detector , whether due to stray light or any other instrumental effect .
the difference image reveals no residual stray light near bright stars [ fig .
[ fig : stray ] ( right ) ] and obvious large scale structures [ fig .
[ fig : stray2 ] ] that could arise from the integrated stray light of the many faint sources .
although small scale changes in the instrumental psf between upper and lower halves of the detector are evident at the locations of bright sources .
the power spectrum of the difference image is very similar to the power spectrum of the ( a - b ) image used to assess the instrumental noise ( fig .
[ fig : power ] ) .
therefore we have verified that the observed large - scale fluctuations are not an artifact related to instrumental stray light .
= 1 = 1.0 zodiacal light is the brightest source of foreground emissions at near - ir ( kelsall et al .
1998 ) , but appears to be extremely smooth with relative fluctuations below 0.2% ( abraham et al . 1997 ) .
more recently , pyo et al . (
2012 ) used mid - ir ( 7 - 24 @xmath9 m ) data _ akari _ data to establish upper limits of @xmath72 at the north ecliptic pole .
expected levels for zodiacal light fluctuations in our measurements have been discussed in the context of the spitzer - based measurements ( kamm1,2 and akmm ) and also in matsumoto et al .
( 2011 ) for the _ akari_-based results . because the _ spitzer _ orbits with the zodiacal cloud , the zodiacal light intensity and the direction of its gradient will vary with a roughly annual period in each field .
the comparison of power spectra at , and cross - correlation power spectra between , the different epochs also serves as an empirical constraint on the influence of the zodiacal light .
the fields in this study lie at high ecliptic latitude well outside the ecliptic plane where zodi is at its highest . in earlier studies
it was already estimated by us to be much smaller than the detected fluctuations .
the gradient due to zodiacal light would vary as the telescope moves around the sun and hence zodiacal light would contribute different levels at different epochs . then any change in the actual sky brightness or power spectra between three epochs of observations separated by @xmath68 months can be attributed to changes in the zodiacal light .
[ fig : cross ] shows the cross - power spectra between the epochs and the signal in each of the epochs for both fields .
the figure shows that the signal remains the same in each of the epochs and correlates remarkably well between them .
this argues against substantial contribution from either the zodiacal light or detector systematics . to sum up this part of discussion
we find no significant differences between epochs , demonstrating that zodiacal light is not a significant contributor to the observed spatial fluctuations .
furthermore , the relatively constant intensity of the fluctuations from 3.6 to 4.5 @xmath9 m is at odds with the measured zodiacal light spectrum which rises by a factor of 2.5 from 3.6 to 4.5 @xmath9 m .
[ fig : cross ] shows that the three separate epochs of data produce the same fluctuation signal and they all correlate very well with each other .
this directly confirms that the contribution of the time - varying component to the measured fluctuations at 3.6 and 4.5 is small and zodiacal fluctuations can not account for the measurement .
galactic cirrus is another source of potential confusion in this measurement .
however , its flux levels depend on the line of sight column density of the galactic ism and lead to variations by factor of a few ( @xmath73 ) among all the seven fields probed in our studies ( kamm1 , kamm2 and seds ) . on the other hand , the measured fluctuation signal at 3.6 and 4.5 is the same , within the statistical uncertainties , at all locations .
this by itself argues against a significant contribution of the galactic cirrus emission .
contributions of galactic cirrus to the source - subtracted fluctuations at the near - ir wavelengths have been addressed earlier in our spitzer analysis ( kamm1 - 2 and akmm ) and in the _ akari_-based study ( matsumoto et al .
all of the studies reached the conclusion that galactic cirrus contributes well below the fluctuation levels measured there .
in addition , in the _ akari_-based analysis it was demonstrated that the near - ir source - subtracted maps do not correlate with the same maps observed at 100 , which is the most accurate probe of cirrus emission ( arendt et al .
1998 , schlegel et al . 1998 ) . in order to further verify that the galactic cirrus emission can not contribute appreciably to the signal measured at 3.6 and 4.5 we proceed as outlined in kamm1,2 and akmm with some extra steps .
we measure identical fluctuations in both fields in both irac bands despite the fact that the mean cirrus flux varies by factors of a few from the uds to the egs .
the level of the fluctuations also agrees well with our earlier measurements at five additional locations out to angular scales of @xmath8 ( kamm1,kamm2 ) . as these high latitude deep survey fields
are chosen in part based on having low ism column densities , it is no surprise that we see no direct evidence of cirrus in these fields .
cirrus emission at 3.6 and 4.5 @xmath9
m is largely continuum emission from stochastically ( transiently ) heated small dust grains and polycyclic aromatic hydrocarbons ( pahs ) .
an empirical measurement of cirrus emission needs to be made at longer wavelengths , such as 100 @xmath9 m where the emission arises from the bulk of the ism dust ( larger grains at cooler temperatures ) , or at 8 @xmath9 m where the emission is dominated by strong pah emission bands that are closely related to the 3.6 and 4.5 @xmath9 m emission . as in our prior studies , in order to evaluate cirrus contributions we have also examined irac 8 images generated from earlier observing programs carried out while _ spitzer _ was still operating cryogenically .
these data are taken with the same irac instrument , and have similar angular resolution to the 3.6 and 4.5 @xmath9 m data .
the 8 @xmath9 m data are somewhat shallower than the seds observations , as the irac 8 @xmath9 m channel was only in operation during earlier cryogenic observations of the egs and uds fields .
the uds field was covered to an average depth of 0.67 hr under spitzer program i d = 40021 , and the egs filed was coved to a mean depth of 1.44 hr under spitzer program i d = 8 .
the 8 uds and egs images were then masked with the masks from 3.6 and 4.5 @xmath9 m and the power spectra computed .
they are shown in fig .
[ fig : cirrus1 ] .
their amplitudes at large scales are roughly proportional to the mean cirrus intensity predicted ( by extrapolation of 100 @xmath9 m observations ) for these fields .
however , this extrapolation is not precise because of intrinsic variation in the energy spectra of the cirrus .
therefore , these power spectra should be viewed as setting a strict upper limit on the possible cirrus contribution at 8 @xmath9 m .
the 8 fluctuations are also expected to contain contributions from the sources producing the measured signal at 3.6 and 4.5 as well as that from other populations , so assuming that _ all _ of the 8 fluctuations signal is produced by cirrus gives the most conservative assumption about its contribution .
= 0.9 in order to estimate directly whether cirrus provides a significant contaminant of the measured fluctuations , we assume that _ all _ of the measured 8 diffuse emission fluctuation on large angular scales ( where shot - noise is negligible ) is generated by cirrus and then computed the cross - power spectrum expressing its value in terms of the _ coherence _ between channel @xmath38 and channel 4 at 8 , @xmath74 . if cirrus emission contributed all of the channel 4 signal and provided a significant contribution to the measurements at 3.6 and 4.5 the coherence would have to be @xmath75 .
however , the results of the computation are shown in fig .
[ fig : cirrus2 ] and indicate that , even if dominant at 8 , galactic cirrus contributes negligibly to the measured signal at 3.6 and 4.5 .
the coherence of the egs field is largely uncertain at the largest scale 3 - 4 data points , where the cross - power @xmath76 is sometimes _ negative _ , but as mentioned above , the levels of cirrus fluctuations there are expected to be at a level similar to that at smaller scales ( e.g. kashlinsky & odenwald 2000 ) , one expects the level of cirrus flux fluctuations at @xmath0 to be similar to that at the smaller scales .
] and also the uds field shows that cirrus contributes negligibly at the largest scales probed here .
the energy spectrum of the cirrus drops by a factor of @xmath77 from 8 to 4.5 and 3.6 @xmath9 m .
therefore , even if the amplitude of the large - scale power spectra at 8 @xmath9 m is dominated by cirrus , the amplitudes at 3.6 and 4.5 @xmath9 m will be at least @xmath78 times lower . at such levels , it remains possible that the cirrus could account for the large scale power spectra .
yet this is an unlikely scenario because calculation of the _ coherence _ expressed via cross - power spectra , @xmath79 ( fig .
[ fig : cirrus2 ] ) shows a relatively weak correlation between the 8 @xmath9 m and shorter wavelength emission .
this conclusion is in agreement with that of the akari analysis ( matsumoto et al .
2011 ) who demonstrate that the near - ir fluctuation do not correlate with the 100 data for the same location implying a negligible contribution of cirrus to the former .
having established the likely cosmological origin of the measured fluctuations , we now turn to interpreting these results .
there is a statistically significant signal on the remaining source - subtracted cib fluctuations in the spitzer data .
it now extends to angular scales @xmath0 and , as demonstrated in appendix a is consistent with our earlier measurements .
kamm1 have proposed that these cib fluctuations originate at early epochs and kamm3 have discussed the properties of the populations needed to reproduce the measured signal . on the other hand , the high-@xmath7 interpretation has been challenged by cooray et al .
( 2007 ) who suggested that faint acs galaxies at intermediate @xmath7 are responsible for the measured signal at 3.6 ( no results for the 4.5 band irac fluctuations are presented there ) because they claim a drop in power at 3.6 when acs - identified galaxies are blanked out and by thompson et al .
( 2007b ) who proposed that the energy spectrum of the fluctuations is consistent with colors of stellar populations at @xmath80 .
akmm ( see secs .
4.2 , 6.3.3 and figs . 5 , 38 there ) have discussed both of these proposals ( cooray et al .
2007 , thompson et al .
2007b ) in detail and showed that high-@xmath7 interpretation remains the best description of all the previous data .
[ fig : cib_fields ] shows that , within the statistical uncertainties , the fields in this study have the same power spectrum of the source - subtracted fluctuations .
therefore , we averaged the two sets of results to obtain an overall power spectrum describing the cib .
the averages were weighted with the number of fourier elements in each field as described above .
the results of the cib fluctuations after averaging over the individual fields are shown in fig .
[ fig : cib_final ] .
the figure indicates the presence of significant large - scale fluctuations remaining after removing / subtracting the resolved sources .
the measured fluctuation spectrum is made of two components : small scales , @xmath81 , are dominated by the shot - noise produced by the discreteness of the remaining sources .
the isotropy of the measured signal , which is further demonstrated in appendix a for five additional fields from our prior measurements , is consistent with it being of cosmological origin . at the same shot - noise level ,
the measured signal is in excellent agreement with our measurements at five other sky locations ( kamm1 - 2 ) at smaller angular scales ( @xmath82 ) .
this section discusses the constraints on the populations producing both the shot - noise and clustering components .
it is organized as follows : we first address the limitations on the fluxes of the individual sources producing the large - scale clustering component which stem from the measured shot - noise levels and galaxy counts .
next we revisit , in light of the new results , the existing estimates of the levels of the clustering component of the cib fluctuations from the remaining known galaxy populations and emphasize , again , that the known galaxy populations do not account for the large - scale clustering component , both its amplitude and spatial dependence . finally , we discuss the limitations on the nature and epochs of the new populations implied by both the measured shot - noise levels and the clustering component of the source - subtracted cib fluctuations . on scales greater than @xmath83 the fluctuation spectrum is dominated by a component due to clustering of the sources producing these cib fluctuations . on scales from @xmath84
to @xmath85 the amplitude of this component remains roughly constant at @xmath86 nw@xmath15/m@xmath87/sr@xmath15 at 3.6 and @xmath88 nw@xmath15/m@xmath87/sr@xmath15 at 4.5 .
the shot noise of the sources that produce these large scale fluctuations can not exceed the observed shot noise component that dominates the fluctuations at scales @xmath89 .
the shot noise level is measured to be @xmath90 njy@xmath91nw / m@xmath15/sr at 3.6 , 4.5 respectively .
these levels can be compared to the shot noise derived from the observed galaxy counts in the fields to set lower limits on the magnitudes ( upper limits on the flux densities ) of the sources that remain unmodeled and unmasked in the background .
galaxy counts were derived independently from the reduction for the background fluctuation analysis .
the _ spitzer_/irac seds data for the egs and uds fields were reduced independently to mosaic form using a combination of facility pipeline processing tools and iracproc ( schuster et al .
the resulting mosaics were constructed with 0.6@xmath92 pixels using temporal outlier rejection to eliminate cosmic rays and instrumental artifacts .
we performed source extraction using sextractor ( betin & arnouts 1996 ) in both fields and both bands a total of four mosaics .
this was done in dual - image mode , using the 3.6@xmath9 m mosaic as the detection image and then carrying out photometry only at the positions of detected sources in the mosaics . to account for source confusion we inserted 10,000 randomly - placed point sources in the mosaics and then attempted to recover them using sextractor with identical parameter settings as
were used for the original source detection . in this way the source counts which are dominated by galaxies below 16 vega mag could be corrected for completeness . for more details see ashby ( 2012 ) , in preparation .
the derived galaxy counts , @xmath93 , were integrated from a lower magnitude limit to infinity to derive the shot noise from the faint galaxies that may remain in the background , @xmath94 .
[ fig : sn_counts ] shows the shot noise levels from the measured counts as a function of the limiting magnitude , @xmath95 .
the shot - noise levels shown in fig .
[ fig : sn_counts ] are in excellent agreement with our earlier such determinations as shown in fig . 1 of kamm3 .
note that the appropriate comparison there should be made to the shot noise levels reached in kamm1 using comparable integrations ; our analysis of goods data ( kamm2 ) with 25 hr / pixel total integrations reached shot noise levels a factor of @xmath96 lower .
consistent with kamm1 - 3 we conclude that the shot - noise level here implies that only sources at @xmath97 remain in the cleaned maps to contribute to the source - subtracted cib clustering component at scales @xmath98 .
this shot noise calculation will suffer from incompleteness in the galaxy counts . at faint magnitudes , @xmath99 , the fields are confusion - limited ( fazio et al .
2004b ) at the irac beam resolution of order @xmath100 .
correcting for incompleteness in this heavy confusion - limit leads to correction factors of @xmath101 and , as discussed in fazio et al .
( 2004b ) , such corrections are suspect when corrections factors exceed @xmath102 .
given this uncertainty , the limiting magnitude is likely to increase to @xmath103 after correcting for incompleteness as discussed in kamm2 .
this is supported by helgason et al ( 2012 ) , who reconstruct galaxy counts from multi - wavelength luminosity function data and derive similar relations between the shot noise levels and the magnitude limits , @xmath95 .
= 0.7 are the sources contributing the clustered component just below the threshold of our source - removal , @xmath104 , or is the clustered component produced by sources significantly fainter with @xmath105 ?
the shot - noise can be expressed as @xmath106 where @xmath107 is the typical flux of the sources contributing to the measured fluctuations and producing the cib of net flux @xmath108 ( kamm3 ) .
thus this can be addressed by comparison between the observed level of the large scale fluctuations , @xmath109 , and the mean cib level implied by the shot noise , @xmath110 , where @xmath111jy is the typical flux of the population producing the large scale clustering component of the cib . if both @xmath112 and @xmath109 are dominated by the brightest sources that were not masked or modeled , @xmath113 , then @xmath114 at 3.6 , 4.5 @xmath9 m all the way to at least @xmath115degree scales .
thus , sources around @xmath116 would need to be essentially completely clustered in structures ( and voids ) with scales of @xmath117 .
this strong large scale clustering is inconsistent with the observed clustering of brighter sources and the predicted clustering of faint normal galaxy populations .
since the shot noise level alone is consistent with normal galaxies at @xmath118 , we conclude that the large scale structure must arise from sources well below our threshold magnitude . to date
there are three estimates of the remaining known galaxy populations to the measured source - subtracted cib fluctuations ( kamm1 , sullivan et al 2007 , helgason et al 2012 ) .
all of these estimates are consistent with each other and are discussed below .
kamm1 have estimated the _ upper _
limit on the cib fluctuations due to the remaining known galaxies as follows .
they estimated the effective limiting magnitude of the remaining galaxies and used it to evaluate net cib flux expected from the known galaxy populations using the counts of fazio et al ( 2004 ) .
then the upper limit on the residual cib fluctuations were estimated by assuming that at all earlier epochs the remaining galaxies had the same clustering pattern as observed today on small scales with the 2-point correlation function @xmath119 .
this upper limit on the clustering component of the remaining cib fluctuations is shown in fig .
[ fig : cib_final ] after convolving with the irac beam as per fig.1 of kamm1 . in a different kind of analysis sullivan
et al ( 2007 ) have estimated the remaining cib fluctuations at 3.6 for the kamm1 parameters by reconstructing the counts in shallow , but wide - field , irac measurements as well as deep goods observations , encompassing the angular scales of kamm1 .
they measured the clustering of resolved sources out to @xmath120 down to ab magnitude of @xmath121 and then used a halo model ( cooray & sheth 2002 ) combined with conditional luminosity functions to estimate the cib fluctuations at 3.6 from sources just below the threshold of the kamm1 analysis .
their results are shown in fig .
[ fig : cib_final ] and are taken from fig .
8 of sullivan et al ( 2007 ) .
a different and substantially more extensive analysis was recently performed in helgason et al ( 2012 ) .
they modelled the remaining cib fluctuations using a massive compilation of 230 datasets of luminosity functions encompassing uv , optical and near - ir bands and spanning a wide range of redshifts . using these data they were able to reconstruct empirically the evolution of the observed populations populating the redshift cone out to @xmath122 at all wavelengths .
the only uncertainties were due to the assumed extremes of the faint end of the luminosity function end , termed high - faint - end ( hfe ) and low - faint - end ( lfe ) .
the counts from these reconstructed populations were evaluated in the observer rest frames and were shown to be in excellent agreement with the measurements down to the faintest measured magnitude at all wavelengths from 0.45 to 4.5 . the shot - noise reconstructed from such data is immune to confusion .
this , coupled with the concordance cosmological model ( @xmath4cdm ) , allowed to robustly evaluate the source - subtracted cib fluctuations at 3.6 and 4.5 due to the populations below the measured shot - noise levels for the hfe and lfe extremes .
these are shown by shaded areas in fig .
[ fig : cib_final ] . fig .
[ fig : cib_final ] shows the power spectrum expected for known galaxies from sullivan et al .
( 2007 ) and helgason et al .
( 2012 ) that are at fainter magnitudes than the shot noise constraint .
this contribution is significantly smaller than the measured level of the fluctuations at scales greater than @xmath83 and also has a very different power spectrum .
all of the estimates are mutually consistent and show that the existing galaxy populations remaining below the measured shot - noise levels can not account for the clustering component of the cib fluctuations measured by us .
the above discussion strongly suggests that the measured cib fluctuations are produced by sources significantly fainter than the magnitude threshold of our source - removal procedure . in this case , it is likely that the measured shot - noise levels represent an _ upper _ limit on the shot - noise from these sources .
no such faint populations have been observed in the present universe and , in addition , if they exist at low-@xmath7 these very faint cosmological sources would have to cluster very differently from the observed galaxy populations which would likely put them in conflict with the established @xmath4cdm cosmological model . on the other hand
, such faint populations are expected at high-@xmath7 epochs , but are beyond the sensitivity and/or resolution of current telescopic studies .
these populations would then be clustered according to the standard @xmath4cdm model except their clustering properties would be biased because they form in haloes identified with rare density peaks of the ( overall ) linear density field at these early times ( kashlinsky et al .
2004 , cooray et al .
2004 , fernandez et al .
2011 ) . to see if such populations can provide a reasonable fit to the measured large - scale clustering we model their power spectrum template as @xmath123 where @xmath124 is a high-@xmath7 @xmath4cdm 3-dimensional template and @xmath125 is to - be - fitted amplitude at the fiducial scale of @xmath126 and @xmath127
is the comoving angular diameter distance to @xmath7 .
we emphasize the toy " nature of such a model : it assumes 1 ) that all the emission occurs over a narrow instant in time , and 2 ) the biasing is linear so that the 3-dimensional power spectrum of emitters is proportional to @xmath124 .
while the second assumption is fairly justifiable on the relevant linear scales , the first supposition is likely to be a significant over - simplification and in reality one needs to relate the projected angular power spectrum to the underlying 3-d one via the limber equation after specifying the history and rate of flux production over a wide range of wavelengths .
thus within the framework of this toy " model , the assumed templates are shown in fig .
[ fig : lcdm ] and are practically independent of @xmath7 at the high redshifts .
this is a consequence of the turn - over scale of the power spectra , which reflects the horizon at the matter - radiation equation in cdm - type models , subtending essentially the same angular scale at @xmath128 . in detailed comparison of model and data , the effect of the masking on the power spectrum
must be considered .
more specifically , the multiplication of the source subtracted images by a mask ( values of 1 or 0 ) , corresponds to a convolution of the power spectrum by the fourier transform of the mask .
the effect would be redistributing some of the power from spatial scales where it is stronger to those where it is weaker .
this has little impact on a shot noise ( and any other white ) spectrum , but can modify the appearance of a spectrum that rises sharply and has features of interest at large spatial scales . thus the @xmath4cdm - type spectrum is affected more significantly than the spectrum of known galaxies .
this is addressed in detail in appendix b. because the spectrum of the fluctuations is measured from a masked map , the model has to be processed correspondingly .
more specifically , in fourier space the fourier transform of the model gets convolved with that of the mask and since the latter is known from our data , this transformation is unique .
this processing is described in appendix b , where fig .
[ fig : masking ] shows the fluctuation spectra after being processed through the mask .
convolution with the mask does not affect the shot - noise component , and affects the spectrum of known galaxies only marginally . however , the @xmath4cdm - type spectrum is affected more significantly by the transfer of power due to mask if it arises at ( high ) redshifts such that the subtended comoving scales reflect its steeply changing part .
( some of the large - scale power will then be transferred to small scales @xmath81 , but this is unobservable since it is overwhelmed by the much larger shot - noise power on these scales . ) the blue solid lines in fig .
[ fig : cib_final ] show the fits to the data over scales @xmath129 with the effect of the masking included .
the corresponding amplitudes of the model are @xmath125 = [ 0.07,0.05 ] nw / m@xmath15/sr at [ 3,6 , 4.5 ] @xmath9 m respectively .
the sum of all three components ( shot - noise plus known galaxies from sullivan et al .
( 2007 ) plus the high-@xmath7 @xmath4cdm population ) is shown in fig .
[ fig : cib_final ] with the red solid line .
the detected source - subtracted fluctuations appear of cosmological origin , but significantly exceed the signal from the observed known galaxy populations remaining in the data after the source - subtraction .
the measured fluctuations are also in excellent agreement with our earlier measurements on the relevant scales and at the same level of shot - noise ( kamm1 - 2 ) .
their spatial spectral distribution on sub - degree scales is also very different and is supportive of these fluctuations originating from sources located at early times and coincidental with the first stars era " .
these sources are individually inaccessible to current telescopic studies and this measurement allows for unique characterization of their properties and spatial distribution . an alternative to this high-@xmath7 interpretation would require the individual sources to be very faint as constrained by the observed shot - noise levels and the absence of correlations with the acs sources ( kamm3 ) , lie at low @xmath7 and have clustering properties significantly different from those of observed galaxy populations .
no such a new population has been observed , or proposed on theoretical grounds , but if true this would represent an important discovery in its own right .
the high-@xmath7 interpretation could be ruled out using optical and shorter wavelength ir imaging such as would be available from cross - correlating our maps with the hst wfc3 and acs images .
the presence of correlations between our irac source - subtracted maps and those in the optical would argue for the new low-@xmath7 populations .
implications of the high-@xmath7 interpretation of the fluctuations have been discussed in kamm3 and we briefly reiterate them here in light of the new measurements .
the typical biasing expected for first haloes in the standard cosmology would lead to relative cib fluctuations of at most @xmath130 on arcminute scales .
thus the measured fluctuations of less than @xmath131 nw / m@xmath15/sr would require the net cib flux at 3.5 and 4.5 of @xmath132 nw / m@xmath15/sr .
such fluxes are well below the claimed mean cib excess levels from the irts and dirbe measurements around 3 ( dwek & arendt 1998 , matsumoto et al . 2005 ) whose theoretical implications have been addressed in santos et al .
( 2002 ) , salvaterra & ferrara ( 2003 ) and madau & silk ( 2005 ) among others .
these are well within the current limits on mean cib from the @xmath133-ray absorption measurements in low-@xmath7 blazars ( dwek et al .
2005 , aharonian et al .
2005 ) , although potentially this contribution may be detectable in the high-@xmath7 grb spectra observed with
lat ( kashlinsky 2005b , kashlinsky & band 2007 , gilmore 2011 ) .
these fluxes are also well within the uncertainties of cib limits from hst deep observations of thompson et al .
( 2007a ) .
kamm3 argue that these populations , having surface density of @xmath134 , must lie in the confusion noise of the present - day space - telescopes , and so care must be made when using filtering in assembling data from individual noisy exposures which may wash out these populations together with the unwanted noise ( akmm ) .
given the short cosmic time available at these epochs to produce the required cib levels these sources must emit radiation much more efficiently with much lower @xmath135 than the present - day stellar populations ( kamm3 ) although there may be some admixture of lower - mass stars ( salvaterra et al .
even further , fluctuation measurements such as this do not directly require the sources of this emission to be exclusively stars and may contain additional contribution from the black - hole accretion emissions in the early universe .
this material is based upon work supported by the nasa adp and hst - c18 grants .
we thank kari helgason for many useful discussions concerning the contributions from the observed galaxy populations to the measured signal . _
facilities : _
unlike galactic or solar system foregrounds , diffuse light fluctuations due to sources at cosmological distances should be isotropically distributed on the sky .
the signal should then be similar at all fields provided the foreground sources are removed to approximately the same shot - noise levels .
[ fig : comparison ] compares all the measurements obtained by us at comparable shot - noise levels ( to within @xmath136 ) decrease with model iteration . in these earlier analyses we used coarser grid of saved iterations than here , so we chose the shot - noise closest to the one reached in this study : @xmath137 nw@xmath15/m@xmath87/sr at ( 3.6 , 4.5 ) . ] .
for historical reasons , and because of its shallower exposure , the field studied in kamm1 at 4.5 had populations removed only to much larger shot - noise level of @xmath138 nw@xmath15/m@xmath87/sr and hence is not shown in the right panel of the figure ; however , its consistency with the kamm2 measurements in the four goods fields at the same shot - noise level is shown in fig . 1 of kamm2 .
the power spectra are evaluated from the masked images .
although in our case the fraction of pixels lost to the mask is small ( @xmath139 ) , it is nonetheless important to evaluate its effects when comparing to predictions of a given input model .
there are three potential contributions to the measured fluctuation spectrum coming from 1 ) shot - noise from remaining sources , 2 ) clustering of known galaxies , and 3 ) from high-@xmath7 sources modelled with @xmath4cdm spectrum .
the galaxy clustering contribution template we adopt per sullivan et al .
( 2007 ) and helgason et al .
the template of the putative high-@xmath7 component of sources with the @xmath4cdm is shown in fig .
[ fig : lcdm ] .
all of these assumed components are measured from a map which is masked after clipping the resolved sources . because the mask is known , this effect can and should be corrected for leading to unique transformation of the assumed template of the power spectrum .
the measured fluctuations are multiplied by the mask template in real angular space .
this is equivalent to convolution in fourier space .
thus the shot - noise will remain flat ( except it is now multiplied by the mask window instead of the beam ) , while the other components with @xmath140const will be convolved in a complicated way with the mask . to understand the effects of the mask and to obtain the templates after mask - processing
we have conducted simulations generating many realizations of a diffuse field of appropriate geometric configuration with a given power spectrum , masking it and computing the resultant power spectrum and its distribution .
[ fig : masking ] shows the effects of the egs and uds masks on the simulated cib spectra of the various components . | we extend previous measurements of cosmic infrared background ( cib ) fluctuations to @xmath0 using new data from the spitzer extended deep survey .
two fields , with depths of @xmath1 hr / pixel over 3 epochs , are analyzed at 3.6 and 4.5 .
maps of the fields were assembled using a self - calibration method uniquely suitable for probing faint diffuse backgrounds .
resolved sources were removed from the maps to a magnitude limit of mag@xmath2 , as indicated by the level of the remaining shot noise .
the maps were then fourier - transformed and their power spectra were evaluated .
instrumental noise was estimated from the time - differenced data , and subtracting this isolates the spatial fluctuations of the actual sky .
the power spectra of the source - subtracted fields remain identical ( within the observational uncertainties ) for the three epochs indicating that zodiacal light contributes negligibly to the fluctuations .
comparing to 8 power spectra shows that galactic cirrus can not account for the fluctuations .
the signal appears isotropically distributed on the sky as required for an extragalactic origin .
the cib fluctuations continue to diverge to @xmath3 times those of known galaxy populations on angular scales out to @xmath0 .
the low shot noise levels remaining in the diffuse maps indicate that the large scale fluctuations arise from the spatial clustering of faint sources well below the confusion noise .
the spatial spectrum of these fluctuations is in reasonable agreement with an origin in populations clustered according to the standard cosmological model ( @xmath4cdm ) at epochs coinciding with the first stars era . #
1= # 1 2sr 2m4sr |
suppose @xmath1 is a density on a state - space @xmath2 with respect to some dominating measure @xmath3 . most often the state - space is a subset of the euclidean space and @xmath3 is the lebesgue measure .
in this article we will always assume @xmath3 to be the lebesgue measure .
statisticians main aim is to study the characteristics of the density @xmath1 .
sometimes ( say , in bayesian inference ) @xmath1 may be a complicated ( possibly unnormalized ) density which is not analytically tractable .
so , to study the characteristics of @xmath1 , statisticians try to draw a sample from @xmath1 .
but then also there may not exist any effective simulation procedure to simulate from @xmath1 .
thus the goal is shifted to draw an approximate sample from @xmath1 .
markov chain monte carlo ( mcmc ) provides a method for doing this .
the mcmc methods , quite famous for their effectiveness in drawing an approximate sample from a target density are widely used .
one of the most famous mcmc algorithms is the metropolis - hastings ( mh ) algorithm @xcite . given a current state @xmath4 the mh algorithm proposes a new state @xmath5 from a _ proposal kernel density _
@xmath6 on @xmath2 , and accepts it with the acceptance probability @xmath7 the corresponding mh kernel has stationary distribution @xmath8 .
the random walk mh ( rwmh ) has the proposal kernel @xmath9 .
we notice that generation a point from such a density is same as generating an @xmath10 from @xmath11 and then setting @xmath12 .
other algorithms like langevin mh ( lmh ) , has proposal kernel @xmath13 all of the aforementioned algorithms have certain disadvantages .
for example , both the rwmh and lmh have slow mixing rates in certain cases .
the acceptance rate for rwmh is high if @xmath14 is concentrated around zero ( i.e. _ small step size _ ) but then a long chain is required to explore a substantial part of the state space . if a diffused proposal is used then the acceptance rate drops . for
a multi modal target ( which is not known a priori in most cases ) both the rwmh and the lmh chain may remain stuck at one or few of the modes and may still pass convergence diagnostics . obviously any inference based on such samples will be incorrect .
also , an important property like geometric ergodicity which are sufficient for clt type results of ergodic averages are either not satisfied by these algorithms ( see * ? ? ?
* for examples ) or some strong assumption on @xmath8 is needed . for example geometric ergodicity to hold for rwmh
it is necessary ( and sufficient ) that the target density is _ log - concave in the tail _
a density @xmath1 continuous and positive on @xmath0 is log - concave in the tail if there exists an @xmath15 and some @xmath16 such that @xmath17 and similarly lmh is also geometrically ergodic under a strong assumption given in theorem 4.1 of @xcite .
these conditions are not satisfied for a large class of densities ( e.g. the densities with thick - tails ) .
thus a mh method which allows the proposed state to be far away from the current state and yet has good acceptance rate will be of much use in the statistical computing problems .
it is even better if the algorithm is geometrically ergodic for a class of densities much larger than the classes for which this property is enjoyed by the standard algorithms . in this article
we propose a new mh algorithm based on multiplying a random quantity with the states .
even if the algorithm appears simple we found that it has excellent convergence and mixing properties .
it can explore the state space quite faster than the standard mcmc algorithms and has geometric ergodicity property for a huge class of target densities , for which the standard algorithms fails to be geometric ergodic .
the main reason for this is because the dives can be made large or small each with significant probabilities .
if the random multiplier is close to one , then the proposed point will be close to the current state and conversely . in rwmh , however , this proposal can not be controlled easily .
if the step size is chosen large then most of the proposed points would be far away from the current states and if the step size is chosen small then most of the proposed points would be very close to the current states .
there is obviously one issue with this algorithm the origin is an absorbing state .
however , this is not vital since in major practical problems the variables are continuous and the origin has no mass .
thus we can safely remove it from the state space without disturbing the convergence .
we emphasize that the rdmh algorithm exploits the multiplicative group structure of @xmath18 .
the algorithm fails when the origin has positive probability attached to it ( for example , when the state - space is the set of integers ) .
the rwmh agorithm still work in that case .
in other problems such as bayesian testing with point null hypothesis and two - sided alternative , where the target distribution has a continuous part and also has a mass at zero , both rdmh and rwmh fails .
mcmc techniques being extremely popular , the literature is rich with algorithms specialized or generic in nature .
many of them are special cases of mh algorithms with different forms of proposal densities .
we refer to @xcite and @xcite for book length discussions .
the structure of the article is as follows .
we describe the new algorithm in section [ sec : rdmh ] . as claimed already , the algorithm is _ new _ in the sense that it is completely different in concept and in structure from the available multiplicative random walk mh .
this is discussed in details in section [ sec : dissimilarity ] .
we discuss its convergence properties in section [ sec : convergence ] . in section [ sec : appl : multimodal ] we compare rdmh with the standard algorithms . specifically we consider a bimodal target and see how rdmh explores the modes while rwmh can not .
we also consider an extreme mixture example where one of the components has very low dispersion compared to other .
we then consider a thick tailed target for which rwmh is _ not _ geometric ergodic while rdmh _
is_. in this example we see how asymptotic normality holds for the ergodic averages using rdmh while it fails to hold for the rwmh algorithm.in section [ sec : appli : pricedata ] we analyze a share price return data .
typically in such problems the posterior of one or more parameters are thick - tailed and asymmetric .
the data and mode we consider are analyzed in @xcite using gibbs sampler .
we found that the gibbs sampler failed to explore the tail of the posterior of the location parameter while rdmh sampler did that with ease .
we conclude this article with an outlook on further works in section [ sec : further ] .
suppose that @xmath1 is a target density function , probably unnormalized , on @xmath19 .
at each iteration the algorithm proposes a state @xmath5 from the current state @xmath20 by multiplying a random quantity @xmath10 with @xmath20 ( i.e. @xmath21 ) .
the proposal is accepted with some probability depending on @xmath20 and @xmath5 .
we can classify the proposals into two classes depending on whether @xmath22 or @xmath23 .
we call the case where @xmath22 an _ inner dive _ and the case where @xmath24 an _ outer dive_. notice that an outer dive can also be obtained by dividing the state @xmath20 by an @xmath10 with @xmath25
. hence we can restrict the set from which the random multiplier @xmath10 is drawn to the set @xmath26 .
obviously the point zero is not considered so that the chain does not get stuck at zero . at each iteration
we can take an inner dive or an outer dive at random .
we call the chain symmetric if the probability for an inner dive is half and asymmetric otherwise . in this article
we shall only consider the symmetric rdmh only .
so with a proposal density @xmath27 for @xmath10 on @xmath28 , the algorithm is given in algorithm [ algo : unid ] .
[ algo : unid ] ' '' '' height 1pt width random dive mh on @xmath19 ' '' '' height 1pt width * input : initial value @xmath29 , and number of iterations @xmath30 .
* for @xmath31 1 .
generate @xmath32 and @xmath33 u@xmath34 independently 2 .
if @xmath35 , set @xmath36 3 .
else set @xmath37 4 .
set @xmath38 * end for ' '' '' height 1pt width notice that rdmh _ is _ an mh algorithm with @xmath39 hence it follows that @xmath8 is indeed stationary for the chain .
however other properties do not follow easily the assumptions in general results discussed in @xcite do not hold in this case . we prove them separately in the following section .
notice also that the terms @xmath40 or its inverse in the acceptance ratios correspond to the jacobian of the transformations : @xmath41 and @xmath42 respectively .
the acceptance ratios are free from the proposal density @xmath43 as in rwmh . for each @xmath44
we define the inner and outer acceptance regions respectively as , @xmath45 let @xmath46 and @xmath47 be the potential inner and outer rejection regions respectively .
we see that for each @xmath48 , the rejection probability is given by @xmath49 substituting @xmath50 in the first integral and @xmath51 in the second we get , @xmath52 obviously , in any mh algorithm the proposal density plays an important role in terms of convergence . in this case also a good choice of @xmath43 is needed for faster convergence .
however as we shall see in theorem [ thm : geoerg ] that the chain is geometric ergodic under an extremely weak restriction on @xmath43 .
hence the discussion on the choices of @xmath43 is postponed till the end of section [ sec : convergence ] .
it is quite straightforward to extend the algorithm to higher dimensions .
the variables may be updated either sequentially or jointly .
while updating jointly at each iteration , outer dives should be applied to a random number of components ( which may be zero ) and inner dives to the rest .
the jacobian terms in the acceptance ratios will then be ratios of products of @xmath10 s .
the algorithm is given in algorithm [ algo : highd ] of section [ sec : further ] .
it is already seen that the rdmh algorithm _ is _ a special case of mh class of algorithms . however , it is _ not _ in any case similar to the random walk type algorithms .
the term _ multiplicative random walk _ ( also known as the log - random walk mh ) is not new in the statistics literature .
however , it has been developed only when the state space is the positive half of the real line as @xmath53 where @xmath54
are i.i.d following some distributions on the real line ( see , dellaportas and roberts , 2003 , pp .
18 for details and jasra _ et .
al . _ , 2005 , for application ) .
obviously this reduces to the simple rwmh on @xmath19 by observing that @xmath55 .
it is useless for problems having entire real line as support because a part of the state space is never visited ( i.e the chain becomes _ reducible _ ) , that is , positive ( negative ) initial values restrict the chain to take only positive ( respectively negative ) values .
the rdmh , however , is developed when the state sapce is entire real line .
the term @xmath56 above can not be equal to @xmath57 since @xmath57 can take both positive and negative values .
hence the rdmh algorithm can not be considered as a special case of log - random walk mh . for distributions with @xmath58 as support
the obvious way to emply rdmh is to reparametrize by taking logarithm so that the support becomes @xmath19 .
let us denote the kernel the of the rdmh chain by @xmath59 .
important properties like irreducibility and aperiodicity ( see * ? ? ?
* for definitions ) are satisfied by the rdmh under minor assumption : [ thm : ergodicity ] if for all @xmath60 , @xmath61 and @xmath8 is bounded and positive on every compact subset of @xmath19 , then the chain is @xmath62-irreducible ( and hence @xmath1-irreducible ) and aperiodic .
suppose @xmath63 and @xmath64 has @xmath65 .
then there exists a compact set @xmath66 , such that @xmath67 where @xmath68 .
define , @xmath69 \qquad a^ * = a{\cap}c \qquad m = \inf_{y\in c } \pi(y ) \quad \textrm { and } \quad m = \sup_{y\in c}\pi(y)\ ] ] and @xmath70 .
the kernel @xmath71 of the chain satisfies , @xmath72 where @xmath73 thus we see that , the chain is @xmath62-_irreducible_. further since each measurable set with positive lebesgue measure can be accessed in a single step implies the chain is _
thus the rdmh chain is _ ergodic _ and hence by theorem 4 of @xcite , @xmath74 for @xmath75almost every @xmath76 .
where , @xmath77 is the well - known _ total variation distance _ between two probability measures @xmath78 and @xmath79 , defined as @xmath80 for further properties of the total variation distance see @xcite or @xcite . a sufficient condition for to hold for all @xmath20 rather than @xmath1almost every @xmath20 is the _
harris recurrence _ of the chain .
a markov chain @xmath81 is called harris recurrent if for every @xmath20 in the state space and for every set @xmath82 such that @xmath83 , @xmath84 = 1.\ ] ] obviously the point _
zero _ creates a problem in our rdmh algorithm .
however , if we remove the single point zero from the state space , then since the chain is already @xmath1-irreducible we use lemma 7.3 of @xcite to conclude that under the assumptions of theorem [ thm : ergodicity ] , the rdmh chain is harris recurrent on @xmath85 . thus holds for any nonzero @xmath20 , i.e. any nonzero starting value ensures convergence of the chain .
a subset @xmath86 of @xmath2 is called _ small _ if there exists a positive integer @xmath87 , a number @xmath88 and a nontrivial measure @xmath89 such that @xmath90 we will characterize the small sets for rdmh . in most of the mh algorithms
any bounded subset of the state space is small .
however this is not the case with rdmh .
we first state a result for rdmh kernel useful in characterizing the small sets .
[ lemma : weakconv ] suppose @xmath91 is a sequence of positive ( negative ) numbers decreasing ( resp . increasing ) to zero , then @xmath92 , where @xmath93 is the distribution degenerated at zero . without loss
we assume @xmath94 .
suppose @xmath95 .
then @xmath96 ) ~\leq ~\dfrac{1}{2}\int_{x_n / y}^0 g({\ensuremath{\epsilon}})d{\ensuremath{\epsilon}}~\to~ 0.\ ] ] also for @xmath97 , for sufficiently large @xmath87 , @xmath98 so that @xmath99 ) \to 1 $ ] .
this completes the proof .
[ thm : smallset ] suppose the conditions in theorem [ thm : ergodicity ] holds .
then a set @xmath100 is small _ if and only if _ its closure , @xmath101 is a compact subset of @xmath102 .
suppose first that @xmath101 is compact subset of @xmath103 .
then @xmath104 so letting @xmath105 , it is seen from the proof of theorem [ thm : ergodicity ] that , @xmath106 since @xmath107 is a nonzero measure on @xmath108 , this shows that @xmath100 is small .
now suppose that @xmath100 is small
. clearly @xmath109 can not be limit points of @xmath100 since for any fixed @xmath87 and bounded @xmath110 , @xmath111 as @xmath112 .
we shall also show that zero can not be a limit point of @xmath100 .
this will show that @xmath101 is compact subset of @xmath103 .
so suppose on the contrary that zero is a limit point of @xmath100
. then there exists a sequence @xmath91 in @xmath100 which monotonically converges to zero .
hence for any @xmath113 and any measurable set @xmath110 , by lemma [ lemma : weakconv ] , @xmath114 so that can not hold for all @xmath115 and all @xmath110 contradicting the assumption that @xmath100 is small .
we now turn towards geometric ergodicity .
an irreducible markov kernel @xmath71 ( irreducible with respect to some @xmath116finite measure @xmath89 ) with invariant distribution @xmath8 is said to be geometric ergodic if @xmath117 for some @xmath118 , where @xmath119 , for @xmath75a.e .
@xmath76 .
geometric ergodicity is important in mcmc applications for the clt of ergodic averages @xmath120 of some function @xmath121 evaluated at each state of the markov chain @xmath122 .
corollary 2.1 of @xcite ( based on work of * ? ? ?
* ) states that if a markov kernel @xmath123 is geometric ergodic and reversible then for any function @xmath121 on the state space such that @xmath124 @xmath125 such a clt easily may not hold if the kernel is not geometric ergodic ( see * ? ? ?
* for examples ) or section [ sec : appli : thick ] of this article .
geometric ergodicity has multifarious usefulness discussed in @xcite and @xcite . to show that rdmh chain is geometrically ergodic , we put the following restriction on @xmath1 . + * assumption ( a1 )
* for some @xmath126 @xmath127 where the notation @xmath128 should be interpreted as 0 for each @xmath129 .
notice that this is true for most of the posterior densities in bayesian literature where mcmc finds extremely high applications .
the condition ( a1 ) given above is basically a regularly varying type restriction on @xmath1 .
further discussion on regularly varying functions can be found in @xcite .
recall that for each @xmath48 , the rejection probability is @xmath130 we now give a bound on @xmath131 in the following lemma .
[ lemma : rhox ] assume ( a1 ) .
then @xmath132 notice that as @xmath133 , @xmath134 and as @xmath135 , @xmath136 .
hence the result follows from dominated convergence theorem .
now we state a helpful result without proof .
[ lemma : silly ] fix @xmath137 . for each @xmath129 and @xmath138
define @xmath139 with @xmath140 .
then * @xmath141 for all @xmath129 and @xmath138 . * @xmath142 for all @xmath129 and @xmath143 .
as discussed in theorem 15.0.1 of @xcite , the rdmh chain is geometric ergodic if and only if for some small set @xmath100 , some function @xmath144 which is finite at least for one @xmath20 , @xmath145 and some @xmath146 , the _ geometric drift condition _ holds : @xmath147 where @xmath148 . in our case , we know any compact set of @xmath103 is small .
so if we can show for some continuous @xmath144 which is bounded on every compact subset of @xmath103 , the following conditions hold : @xmath149 then we can choose a number @xmath145 and a small set @xmath150 for some @xmath151 , such that @xmath152 for all @xmath153 .
also @xmath154 since @xmath155 is bounded on @xmath100 .
hence we see that holds .
we now state and prove the most important theorem of this section . [ thm : geoerg ] suppose that conditions in theorem [ thm : ergodicity ] holds together with continuity of @xmath156 and ( a1 ) .
further assume the following : for some @xmath157 , @xmath158 then the chain is geometrically ergodic . in view of discussion preceding the statement of the theorem we only need to show .
fix @xmath159 .
then by @xmath160 now notice that , for each @xmath48 , and any function @xmath161 @xmath162 choose positive constants @xmath163 , @xmath164 and @xmath165 such that the function @xmath166 is continuous , @xmath167 and @xmath168 and @xmath169 for all @xmath48 and let @xmath170 and @xmath171 we now work with this @xmath155 to show suppose @xmath172 . then assumption ( a1 ) implies that @xmath173 and @xmath174 . notice in this case @xmath175 hence by the second integral in converges to zero . also , since @xmath176 @xmath177 where @xmath178 , the third integral in also converges to zero .
further since @xmath179 it follows from and lemma [ lemma : rhox ] that @xmath180 suppose now that @xmath181 . in this case @xmath182 and @xmath183 .
also @xmath184 for all @xmath185 implies that the first integral in converges to zero by . also @xmath186 and @xmath187 @xmath188 where @xmath189 , so that the fourth integral in converges to zero .
hence by continuity of @xmath8 at zero , @xmath190 this completes the proof .
* remark : * it is conjectured in @xcite that an mh chain is geometric ergodic if the rejection probability is bounded away from 1 .
they have proved the result with an additional assumption that the continuous part of the mh kernel , i.e. which is an operator on @xmath191 , is compact .
unfortunately this extra assumption does not hold for rdmh ( along with most of the mh algorithms ) .
had the conjecture been proved , we could have claimed readily that rdmh is geometrically ergodic by using lemma [ lemma : rhox ] .
this would not require the extra assumption on @xmath43 .
the class of densities satisfying ( a1 ) together with the assumptions of theorem 1 and continuity is quite large .
this class obviously includes the following classes : 1 .
the class of thick - tailed densities @xmath192 as @xmath112 where @xmath193 is a polynomial satisfying @xmath194 for all sufficiently large @xmath195 . for example , the @xmath196-densities fall in this class .
the class of densities which are equally log - concave in the two tails , i.e. , for some @xmath197 and some @xmath15 , @xmath198 this is a stronger version of for implies .
notice that for these densities , @xmath199 in ( a1 ) .
examples of such densities are the normal densities and their mixtures , double exponential density etc .
3 . the class of densities of the form @xmath200{|x-\theta|})$ ] where @xmath201 . in some problems , however , the target @xmath156 is log - concave in the tails but the rates at which @xmath156 converges to zero are not same for the two tails .
one example of such densities is @xmath202 .
notice that if @xmath203 follows the standard exponential distribution @xmath204 then @xmath205 has density @xmath206 .
it can be seen that @xmath207 holds for each @xmath208 , as @xmath112 and also for each @xmath209 and as @xmath210 .
but @xmath211 if @xmath209 and @xmath212 .
we assume for these kind of densities exactly one tail dominates , i.e. exactly one the following is true for each @xmath213 .
@xmath214 notice that it is sufficient to work with because if holds for a target @xmath1 , then @xmath215 satisfies . for these kind of densities ( a1 ) does not hold .
however , the next theorem assures that the rdmh chain is still geometric ergodic .
the proof is along the line of theorem [ thm : geoerg ] and so we just present a sketch .
suppose @xmath1 is continuous and the assumptions in theorem [ thm : ergodicity ] holds .
suppose further that @xmath156 satisfies and that @xmath43 satisfies the regularity condition .
then the rdmh chain is geometric ergodic .
notice that in this case we have the following as @xmath210 we still have @xmath216 and @xmath174 .
but @xmath217 and @xmath218 thus in [ eqn : kernratio ] ( with the same choice of @xmath155 as in theorem [ thm : geoerg ] ) we can further split the integrals on intersections of the domains with @xmath219 and @xmath34 . on each such domain
either the integrand converges to zero or it is bounded and the domain of the integral converges to the empty set .
hence @xmath220 since , @xmath221 also as @xmath222 , it can be seen that @xmath173 and @xmath223 hold but @xmath224 and @xmath225 hence similarly , @xmath226 since in this case @xmath227 this verifies the first condition of .
verification of the second condition of is already done in _ case ii _ of theorem [ thm : geoerg ] .
we now return to the choices of @xmath43 .
let @xmath228 denote the density of a beta@xmath229 random variable . a general class of proposal densities satisfying is then given by @xmath230 for some @xmath231 and some positive numbers @xmath232 .
notice that a straightforward choice of @xmath43 is uniform distribution over ( -1,1 ) which corresponds to the case @xmath233 and @xmath234 .
this indeed allows for large dives but the acceptance rate may drop .
further if the simulated @xmath10 is very close to zero and a outer dive is taken , then the proposed state will have large magnitude and result in numerical instability .
specially when the posterior is highly steep then such large dives are not sensible .
we shall nevertheless use this choice of @xmath43 in the next section and show it works well .
if however , the target density is steep then a good idea would be to generate @xmath10 s close to 1 .
this can be achived by 1 . making @xmath235 small ( e.g. @xmath236 ) and 2 . making @xmath237 high and @xmath238 small ( e.g. @xmath239 and @xmath240 ) .
in this section we consider two simulation studies . we first consider a bimodal target density and show how the rdmh algorithm explores the modes but the rwmh chain either gets stuck at the mode ( if the proposal variance is moderate ) or explores the modes at high value of proposal variance but has very low acceptance rate . in the next example we consider another simulation study on a thick tailed target .
the rwmh and the lmh algorithms are _ not _ geometrically ergodic for this target under any kind of proposal ( thick - tailed or thin tailed ) and this has a serious effect when we try to construct a confidence set based on asymptotic normality of ergodic averages for the latter does not hold in this case .
however the rdmh is still geometric ergodic and a clt holds for the ergodic averages .
[ [ example-1 . ] ] example 1 .
+ + + + + + + + + + consider the mixture distribution @xmath241 where @xmath242 is the normal density with mean @xmath243 and variance @xmath244 .
clearly this is a bimodal distribution with two separated modes at @xmath245 and @xmath246 .
we compare the rdmh with the rwmh here .
we choose @xmath43 , as the uniform distribution on @xmath26 i.e. @xmath247 , scaledwidth=90.0% ] \(a ) rwmh : @xmath248 , initial value : @xmath249 acceptance rate = 29.916% , ( b ) rwmh : @xmath248 , initial value : @xmath250 acceptance rate = 29.656% , ( c ) rwmh : @xmath251 , initial value : @xmath250 acceptance rate = 14.31% , ( d ) rdmh , initial value : @xmath252 , acceptance rate = 30.172% [ fig : bimodal ] for the rwmh , we choose the proposal @xmath253 for different choices of @xmath254 . figure [ fig : bimodal ]
( panels ( a ) and ( b ) ) shows that rwmh remains stuck at one of the modes for an arbitrary but reasonable choice of @xmath254 , and with arbitrary initial value .
this indicates significant non - robustness of rwmh with respect to the initial value and the choice of @xmath254 even if it is geometric ergodic in this case . only when @xmath254 has been appropriately chosen , rwmh performs adequately ( figure [ fig : bimodal ] , panel ( c ) ) .
we remark that such right " choice is possible only if bimodality of the target posterior is anticipated beforehand , which is unrealistic . even for the appropriate choice of @xmath254
we notice that the acceptance rate of rwmh is rather small ( 14.31% ) .
in contrast rdmh adequately explored the entire state space without requiring knowledge of the target density ( figure [ fig : bimodal ] , panel ( d ) ) , or tuning of the proposal .
the acceptance rate , which is 30.172% , much encouraging compared to the rwmh algorithm .
[ [ example-2 . ] ] example 2 .
+ + + + + + + + + + now we consider a more challenging case similar to one considered by @xcite in their example 2 .
the target is a mixture of univariate normals : @xmath255 several specialized algorithms are available for such _ needle - in - haystack _
problem among which the equi - energy sampler by @xcite is worth mentioning .
the equi - energy sampler is extremely efficient once the tuning parameters are chosen carefully . for our purpose
we chose few proposals at random and study their performances .
in particular , we run each chain ( of length 30,000 each after discarding first 20,000 burn - ins ) 100 times and estimate @xmath256 , where @xmath257 denotes the markov chain . from table [
tab : needle_example_table ] it can be seen that all the proposals work quite well .
the third and fifth proposal results small m.s.es perhaps due to the fact that they put more weight near zero ( the multiplier is close to zero and hence so is the proposed state ) and one of the mode is at zero .
however , since in practice , it need not be the case it might result in poor acceptance rates .
so , a proposal that generates random multiplier close to 1 should be preferred .
.mean , standard deviation and mean squared errors of @xmath258 for different proposals .
the average acceptance rates are also reported . [ cols="^,^,^,^,^",options="header " , ] @xcite used a gibbs sampler approach with data - augmentation .
they faced some numerical difficulties and perturbed the @xmath259 s slightly to resolve the numerical problems .
the rdmh sampler , however , did not face any numerical problem .
the results obtained by rdmh differs from the gibbs sampler perhaps due to this reason . actually , the posteriors of @xmath260 and @xmath235 were same whether we used gibbs sampler or not ( see the paper by @xcite for the gibbs sampler output ) .
the posterior of @xmath261 were quite dissimilar for the rdmh and gibbs samplers .
for the gibbs sampler the posterior was mainly concentrated between @xmath262 and @xmath263 while it was concentrated between @xmath264 and @xmath265 for the rdmh chain .
clearly , the gibbs sampler fails to cover the long tail of the posterior of @xmath261 while the rdmh explores it quite easily . to ensure we also ran a random walk mh sampler and found that the results for the rwmh sampler coincided with that of the rdmh sampler .
the summaries of the rdmh sampler is given in table [ tab : post_beta_sigma ] and [ tab : post_nu_gam ] and the histograms and traceplots of the same are given in figure [ fig : hist_pricedata ] and [ fig : trace_pricedata ] respectively .
the autocorrelation plots of the rdmh chains are given in figure [ fig : acf_price ] .
we ran the sampler for 160,000 iterations and discarded the first 10,000 samples as burn - ins .
we then thinned the remaining 150,000 samples by 5 .
convergence was achieved much earlier though .
we also found that the mixing for the rdmh sampler was superior to that of the rwmh sampler .
we conclude this article with some purview of possible extension to higher dimension .
suppose @xmath1 is a density supported on @xmath266 and @xmath43 is density on @xmath267 .
then the algorithm is given in algorithm [ algo : highd ] .
[ algo : highd ] ' '' '' height 1pt width random dive mh on @xmath266 ' '' '' height 1pt width * input : initial value @xmath268 with no component equal to 0 , and number of iterations @xmath30 . * for @xmath31 1 .
generate @xmath269 and @xmath270 u@xmath271 independently 2 . for each @xmath272 if @xmath273 , set @xmath274 .
else set @xmath275 3 .
let @xmath276 and set @xmath277 4 .
set @xmath278 * end for ' '' '' height 1pt width this algorithm is still irreducible and aperiodic .
it is also harris recurrent on @xmath279 and every compact subset of @xmath279 is still small .
the proof is along the same line as theorem [ thm : ergodicity ] and [ thm : smallset ] .
geometric ergodicity is , however , a property that requires a different approach .
it is expected that geometric ergodicity of this algorithm still holds for a large class of densities ( especially the thick - tailed ones ) on higher dimensions .
we hope that this article would draw attention of the researchers and the question regarding geometric ergodicity in higher dimension situation would be settled .
the proposal density @xmath280 on @xmath267 can be chosen to be the product of proposal densities on @xmath28 .
in such a case , one should choose the univariate proposals which generate @xmath10 s close to 1 with high probabilities each ( for example , the mixture proposals in section [ sec : appli : pricedata ] ) .
this will ensure that the proposed states are not too far away from the current state ( in @xmath266 ) to reduce the acceptance rate significantly .
thanks are due to two anonymous referees whose critical reading of the manuscript and constructive comments lead to major improvement of the paper . | in this article we propose multiplication based random walk metropolis hastings ( mh ) algorithm on the real line .
we call it the random dive mh ( rdmh ) algorithm .
this algorithm , even if simple to apply , was not studied earlier in markov chain monte carlo literature .
one should not confuse rdmh with rwmh .
it is shown that they are different , conceptually , mathematically and operationally . the kernel associated with therdmh algorithm
is shown to have standard properties like irreducibility , aperiodicity and harris recurrence under some mild assumptions .
these ensure basic convergence ( ergodicity ) of the kernel .
further the kernel is shown to be geometric ergodic for a large class of target densities on @xmath0 .
this class even contains realistic target densities for which random walk or langevin mh are not geometrically ergodic .
three simulation studies are given to demonstrate the mixing property and superiority of rdmh to standard mh algorithms on real line .
a share - price return data is also analyzed and the results are compared with those available in the literature . * key words * : markov chain monte carlo ; metropolis - hastings algorithm ; random walk algorithm ; langevin algorithm ; multiplicative random walk ; geometric ergodicity ; thick tailed density ; share - price return . *
ams classification number * _ primary _ : 65c05 , 65c40 .
_ secondary _ : 60j10 . |
the rare charmless b meson decays arouse more and more interest , since it is a good place for testing the standard model ( sm ) , studying cp violation and looking for possible new physics beyond the sm .
since 1999 , the b factories in kek and slac collect more and more data sample of rare b decays . in the future cern large hadron collider beauty experiments ( lhc - b ) , the heavier @xmath6 and @xmath7 mesons can also be produced . with the bright hope in lhc - b experiments and btev experiments at fermilab , following a previous study of @xmath8 decay @xcite
, we continue to investigate other @xmath6 rare decays .
the most difficult problem in theoretical calculation of non - leptonic @xmath9 decays is the calculation of hadronic matrix element .
the widely used method is the factorization approach ( fa ) @xcite .
it is a great success in explaining the branching ratio of many decays @xcite , although it is a very simple method . in order to improve the theoretical precision , qcd factorization @xcite and perturbative qcd approach ( pqcd ) @xcite are developed .
perturbative qcd factorization theorem for exclusive heavy - meson decays has been proved some time ago , and applied to semi - leptonic @xmath10 decays @xcite , the non - leptonic @xmath11 @xcite , @xmath12 @xcite decays .
pqcd is a method to factorize hard components from a qcd process , which can be treated by perturbation theory .
non - perturbative parts are organized in the form of universal hadron light cone wave functions , which can be extracted from experiments or constrained by lattice calculations and qcd sum rules .
more information about pqcd approach can be found in @xcite . in this paper
, we would like to study the @xmath13 and @xmath1 decays in the perturbative qcd approach . in our calculation
, we ignore the soft final state interaction because there are not many resonances near the energy region of @xmath6 mass .
our theoretical formulas for the decay @xmath14 in pqcd framework are given in the next section . in section [ sc :
neval ] , we give the numerical results of the branching ratio of @xmath14 and discussions for cp asymmetries and the form factor of @xmath15 etc . at
last , we give a short summary in section [ summ ] . decay . ]
for decay @xmath16 , the related effective hamiltonian is given by @xcite @xmath17-v_{tb}^*v_{td}\sum_{i=3}^{10}c_i(\mu ) o_i(\mu)\right\},\label{hami}\ ] ] where @xmath18 are wilson coefficients at the renormalization scale @xmath19 and @xmath20 are the four quark operators @xmath21 here @xmath22 and @xmath23 are @xmath24 color indices ; the sum over @xmath25 runs over the quark fields that are active at the scale @xmath26 , i.e. , @xmath27 .
operators @xmath28 come from tree level interaction , while @xmath29 are qcd - penguins operators and @xmath30 come from electroweak - penguins .
working at the rest frame of @xmath6 meson , we take kaon and pion masses @xmath31 , which are much smaller than @xmath32 . in the light - cone coordinates
, the momenta of the @xmath6 , @xmath33 and @xmath34 can be written as : @xmath35 denoting the light ( anti-)quark momenta in @xmath9 , @xmath33 and @xmath34 as @xmath36 , @xmath37 and @xmath38 , respectively , we can choose : @xmath39 in the following , we start to compute the decay amplitudes of @xmath40 . according to effective hamiltonian ( [ hami ] )
, we draw the lowest order diagrams of @xmath40 in fig .
[ figure : fig1 ] .
let us first look at the usual factorizable diagrams ( a ) and ( b )
. they can give the @xmath41 form factor if take away the wilson coefficients .
the operators @xmath42 and @xmath43 are @xmath44 currents , and the sum of their contributions is given by @xmath45 = 16\pi c_f m_b^2 \int_0 ^ 1\!\!\ !
dx_1 dx_2 \int_0^\infty\!\!\!\!\
! b_1 db_1\ , b_2 db_2\ \phi_b(x_1,b_1 ) \\ \times \bigl\ { [ ( 2-x_{2 } ) \phi_{k}^a(x_2)-r_k(1 - 2x_{2})\phi_{k}^p(x_2)\\ + r_k(1 - 2x_{2})\phi_{k}^t(x_2 ) ] \alpha_{s}(t_{a}^{1})h_{a}(x_{1},1-x_2,b_1,b_2 ) \exp[-s_b(t_{a}^{1})-s_k(t_{a}^{1})]c(t_{a}^{1})\\ + 2r_k\phi_{k}^p(x_2)\alpha_{s}(t_{a}^{2 } ) h_{a}(1-x_{2},x_1,b_2,b_1)\exp[-s_b(t_{a}^{2})-s_k(t_{a}^{2})]c(t_{a}^{2 } ) \bigr\},\label{fe}\end{gathered}\ ] ] where @xmath46 $ ] , @xmath47 $ ] .
@xmath48 is the group factor of the @xmath49 gauge group .
the expressions of the meson distribution amplitudes @xmath50 , the sudakov factor @xmath51 , and the functions @xmath52 are given in the appendix .
in above formula , the wilson coefficients @xmath53 of the corresponding operators are process dependent .
the operator @xmath54 have the structure of @xmath55 , their amplitude is @xmath56 = 32\pi c_f m_b^2 r_{\pi}\int_0 ^ 1\!\!\ ! dx_1 dx_2 \int_0^\infty\!\!\!\!\
! b_1 db_1\ , b_2 db_2\ \phi_b(x_1,b_1 ) \\ \times \bigl\ { [ \phi_{k}^a(x_2)-r_k(x_{2}-3)\phi_{k}^p(x_2)\\ + r_k(1-x_{2 } ) \phi_{k}^t(x_2 ) ] \alpha_{s}(t_{a}^{1})h_{a}(x_{1},1-x_2,b_1,b_2)\exp[-s_b(t_{a}^{1 } ) -s_k(t_{a}^{1})]c(t_{a}^{1})\\ + 2r_k\phi_{k}^p(x_2 ) \alpha_{s}(t_{a}^{2})h_{a}(1-x_{2},x_1,b_2,b_1)\exp[-s_b(t_{a}^{2 } ) -s_k(t_{a}^{2})]c(t_{a}^{2})\bigr\}.\label{fep}\end{gathered}\ ] ] for the non - factorizable diagrams ( c ) and ( d ) , all three meson wave functions are involved . using @xmath57 function @xmath58
, the integration of @xmath59 can be preformed easily . for the @xmath44 operators
the result is : @xmath60 = -\frac{32}{3}\pi c_f\sqrt{2n_{c } } m_b^2 \int_0 ^ 1\!\!\ !
dx_1 dx_2dx_3 \int_0^\infty\!\!\!\!\ !
b_3 db_3\ \phi_b(x_1,b_3 )
\\ \times \bigl\ { [ ( x_3 - 1)\phi_{\pi}^a(x_3)\phi_{k}^a(x_2)+r_k(1-x_{2})\phi_{\pi}^a(x_3 ) \phi_{k}^p(x_2 ) + r_k(1-x_{2})\phi_{\pi}^a(x_3)\phi_{k}^t(x_2)]c(t_c^1)\\ \alpha_{s}(t_{c}^{1})h_{c}^{(1)}(x_{1},x_2,x_3,b_2,b_3)\exp[-s_b(t_{c}^{1 } ) -s_{\pi}(t_{c}^{1})-s_k(t_{c}^{1 } ) ] -[(x_2-x_3 - 1)\phi_{\pi}^a(x_3)\phi_{k}^a(x_2)\\ + r_k(1-x_{2})\phi_{\pi}^a(x_3)\phi_{k}^p(x_2 ) -r_k(1-x_{2})\phi_{\pi}^a(x_3)\phi_{k}^t(x_2)]c(t_c^2)\\ \alpha_{s}(t_{c}^{2})h_{c}^{(2)}(x_{1},x_2,x_3,b_2,b_3)\exp[-s_b(t_{c}^{2 } ) -s_{\pi}(t_{c}^{2})-s_k(t_{c}^{2 } ) ] \bigr\}.\label{me}\end{gathered}\ ] ] for the@xmath55 operators , the formula is : @xmath61 = -\frac{32}{3}\pi c_f\sqrt{2n_{c } } m_b^2 r_{\pi}\int_0 ^ 1\!\!\
! dx_1 dx_2dx_3 \int_0^\infty\!\!\!\!\ !
b_3 db_3\ \phi_b(x_1,b_3 ) \\
\times \bigl\{[r_k(x_{2}+x_3 - 2)\phi_{\pi}^p(x_3)\phi_{k}^p(x_2)-r_k ( x_2-x_3)\phi_{\pi}^p(x_3)\phi_{k}^t(x_2)-r_k(x_2-x_3)\phi_{\pi}^t(x_3 ) \phi_{k}^p(x_2)\\ -r_k(2-x_{2}-x_3)\phi_{\pi}^t(x_3)\phi_{k}^t(x_2)- ( 1-x_3)\phi_{\pi}^p(x_3)\phi_{k}^a(x_2)-(1-x_3)\phi_{\pi}^t(x_3 ) \phi_{k}^a(x_2)]c(t_c^1)\\ \alpha_{s}(t_{c}^{1})h_{c}^{(1)}(x_{1},x_2,x_3,b_2,b_3)\exp[-s_b(t_{c}^{1 } ) -s_{\pi}(t_{c}^{1})-s_k(t_{c}^{1 } ) ] + [ r_k(1-x_{2}+x_3)\phi_{\pi}^p(x_3)\phi_{k}^p(x_2)\\ + r_k(x_2+x_3 - 1)\phi_{\pi}^p(x_3)\phi_{k}^t(x_2 ) -r_k(x_2+x_3 - 1)\phi_{\pi}^t(x_3)\phi_{k}^p(x_2 ) -r_k(1-x_{2}+x_3)\phi_{\pi}^t(x_3)\phi_{k}^t(x_2)\\ + x_3\phi_{\pi}^p(x_3)\phi_{k}^a(x_2)-x_3\phi_{\pi}^t(x_3)\phi_{k}^a(x_2 ) ] c(t_c^2)\\ \alpha_{s}(t_{c}^{2})h_{c}^{(2)}(x_{1},x_2,x_3,b_2,b_3)\exp[-s_b(t_{c}^{2 } ) -s_{\pi}(t_{c}^{2})-s_k(t_{c}^{2 } ) ] \bigr\}.\label{mep}\end{gathered}\ ] ] similar to ( c),(d ) , the annihilation diagrams ( e ) and ( f ) also involve all three meson wave functions .
here we have two kinds of amplitudes , @xmath62 is the contribution containing the operator of type @xmath44 , and @xmath63 is the contribution containing the operator of type @xmath55 . @xmath64 = -\frac{32}{3}\pi c_f\sqrt{2n_{c } } m_b^2 \int_0 ^ 1\!\!\ ! dx_1 dx_2dx_3 \int_0^\infty\!\!\!\!\
! b_1 db_1\
b_2 db_2\ \phi_b(x_1,b_1 ) \\
\times \bigl\{[x_3\phi_{\pi}^a(x_3)\phi_{k}^a(x_2)+r_{\pi}r_k ( 2+x_2+x_3)\phi_{\pi}^p(x_3)\phi_{k}^p(x_2)-r_{\pi}r_k(x_2-x_3)\phi_{\pi}^p(x_3 ) \phi_{k}^t(x_2)\\ -r_{\pi}r_k(x_2-x_3)\phi_{\pi}^t(x_3)\phi_{k}^p(x_2)-r_{\pi}r_k ( 2-x_2-x_3)\phi_{\pi}^t(x_3)\phi_{k}^t(x_2)]c(t_e^1)\\ \alpha_{s}(t_{e}^{1})h_{e}^{(1)}(x_{1},x_2,x_3,b_1,b_2)\exp[-s_b(t_{e}^{1 } ) -s_{\pi}(t_{e}^{1})-s_k(t_{e}^{1 } ) ] -[x_2\phi_{\pi}^a(x_3)\phi_{k}^a(x_2)\\ + r_{\pi}r_k(x_2+x_3)\phi_{\pi}^p(x_3)\phi_{k}^p(x_2 ) + r_{\pi}r_k(x_2-x_3)\phi_{\pi}^p(x_3)\phi_{k}^t(x_2 ) + r_{\pi}r_k(x_2-x_3)\phi_{\pi}^t(x_3)\phi_{k}^p(x_2)\\ + r_{\pi}r_k(x_2+x_3)\phi_{\pi}^t(x_3)\phi_{k}^t(x_2)]c(t_e^2)\\ \alpha_{s}(t_{e}^{2})h_{e}^{(2)}(x_{1},x_2,x_3,b_1,b_2)\exp[-s_b(t_{e}^{2 } ) -s_{\pi}(t_{e}^{2})-s_k(t_{e}^{2 } ) ] \bigr\},\label{ma}\end{gathered}\ ] ] @xmath65 = -\frac{32}{3}\pi c_f\sqrt{2n_{c } } m_b^2 \int_0 ^ 1\!\!\ ! dx_1 dx_2dx_3 \int_0^\infty\!\!\!\!\ ! b_1 db_1\
b_2 db_2\ \phi_b(x_1,b_1 ) \\
\times \bigl\{[r_k(2-x_2)\phi_{\pi}^a(x_3)\phi_{k}^p(x_2)+r_k ( 2-x_2)\phi_{\pi}^a(x_3)\phi_{k}^t(x_2)-r_{\pi}(2-x_3)\phi_{\pi}^p(x_3 ) \phi_{k}^a(x_2)\\ -r_{\pi}(2-x_3)\phi_{\pi}^t(x_3)\phi_{k}^a(x_2)]c(t_e^1)\\ \alpha_{s}(t_{e}^{1})h_{e}^{(1)}(x_{1},x_2,x_3,b_1,b_2)\exp[-s_b(t_{e}^{1 } ) -s_{\pi } ( t_{e}^{1})-s_k(t_{e}^{1 } ) ] + [ r_kx_2\phi_{\pi}^a(x_3)\phi_{k}^p(x_2)\\ + r_kx_2\phi_{\pi}^a(x_3)\phi_{k}^t(x_2 ) -r_{\pi}x_3\phi_{\pi}^p(x_3)\phi_{k}^a(x_2 ) -r_{\pi}x_3\phi_{\pi}^t(x_3)\phi_{k}^a(x_2)]c(t_e^2)\\ \alpha_{s}(t_{e}^{2})h_{e}^{(2)}(x_{1},x_2,x_3,b_1,b_2)\exp[-s_b(t_{e}^{2 } ) -s_{\pi}(t_{e}^{2})-s_k(t_{e}^{2 } ) ] \bigr\}.\label{map}\end{gathered}\ ] ] the factorizable annihilation diagrams ( g ) and ( h ) involve only two light mesons wave functions .
@xmath66 is for @xmath44 type operators , and @xmath67 is for @xmath55 type operators : @xmath68 = 16\pi c_f m_b^2 \int_0 ^ 1\!\!\ !
dx_2 dx_3 \int_0^\infty\!\!\!\!\ !
b_2 db_2\ , b_3 db_3\
\\ \times \bigl\ { [ -x_{2}\phi_{\pi}^a(x_3)\phi_{k}^a(x_2)-2r_{\pi}r_k(1+x_{2})\phi_{\pi}^p(x_3 ) \phi_{k}^p(x_2)+ 2r_{\pi}r_k(1-x_{2})\phi_{\pi}^p(x_3)\phi_{k}^t(x_2)]\\ \alpha_{s}(t_{g}^{1})h_{g}(x_2,x_3,b_2,b_3)\exp[-s_{\pi}(t_{g}^{1})-s_k(t_{g}^{1 } ) ] c(t_g^1)\\ +
[ x_3\phi_{\pi}^a(x_3)\phi_{k}^a(x_2)+2r_{\pi}r_k(1+x_3)\phi_{\pi}^p(x_3 ) \phi_{k}^p(x_2 ) -2r_{\pi}r_k(1-x_3)\phi_{\pi}^t(x_3)\phi_{k}^p(x_2)]\\ c(t_g^2)\alpha_{s}(t_{g}^{2})h_{g}(x_{3},x_2,b_3,b_2 ) \exp[-s_{\pi}(t_{g}^{2})-s_k(t_{g}^{2})]\bigr\},\label{fa}\end{gathered}\ ] ] @xmath69 = 32\pi c_f m_b^2 \int_0 ^ 1\!\!\ !
dx_2 dx_3 \int_0^\infty\!\!\!\!\
! b_2 db_2\ , b_3 db_3\
\\ \times \bigl\ { [ r_kx_{2}\phi_{\pi}^a(x_3)\phi_{k}^p(x_2)-r_kx_{2}\phi_{\pi}^a(x_3 ) \phi_{k}^t(x_2)+ 2r_{\pi}\phi_{\pi}^p(x_3)\phi_{k}^a(x_2)]\\ \alpha_{s}(t_{g}^{1})h_{g}(x_2,x_3,b_2,b_3)\exp[-s_{\pi}(t_{g}^{1})-s_k(t_{g}^{1 } ) ] c(t_g^1)\\ + [ 2r_k\phi_{\pi}^a(x_3)\phi_{k}^p(x_2)+r_{\pi}x_3\phi_{\pi}^p(x_3)\phi_{k}^a(x_2 ) -r_{\pi}x_3\phi_{\pi}^t(x_3)\phi_{k}^a(x_2)]\\ c(t_g^2 ) \alpha_{s}(t_{g}^{2})h_{g}(x_{3},x_2,b_3,b_2 ) \exp[-s_{\pi}(t_{g}^{2})-s_k(t_{g}^{2})]\bigr\}.\label{fap}\end{gathered}\ ] ] from equation ( [ fe])-([fap ] ) , the total decay amplitude for @xmath70 can be written as @xmath71\hspace*{3cm}\nonumber\\ -f_{\pi}v_{tb}^{*}v_{td}f_{e}^{p}\left[\frac{1}{3}c_{5}+c_{6 } + \frac{1}{3}c_{7}+c_{8}\right ] + m_{e}\left[v_{ud}v_{ub}^{*}c_{1}-v_{tb}^{*}v_{td}(c_{3}+c_{9})\right ] \hspace*{1cm}\nonumber\\ -v_{tb}^{*}v_{td}m_e^{p}\left(c_{5}+c_{7}\right)-v_{tb}^{*}v_{td}m_a\left(c_{3 } -\frac{1}{2}c_{9}\right ) -v_{tb}^{*}v_{td}m_a^p\left(c_{5}-\frac{1}{2}c_{7}\right)\hspace*{2cm}\nonumber\\ -f_{b } v_{tb}^{*}v_{td}f_a\left[\frac{1}{3}c_{3}+c_{4}-\frac{1}{6}c_{9 } -\frac{1}{2}c_{10}\right ] -f_{b}v_{tb}^{*}v_{td}f_a^{p}\left[\frac{1}{3}c_{5}+c_{6}-\frac{1}{6}c_{7 } -\frac{1}{2}c_{8}\right ] , \label{eq : width}\end{aligned}\ ] ] and the decay width is expressed as @xmath72 should be calculated at the appropriate scale t which can be found in the appendix of ref .
the decay amplitude of the charge conjugate channel @xmath73 can be obtained by replacing @xmath74 to @xmath75 and @xmath76 to @xmath77 in eq.([eq : width ] ) . for the decay @xmath78 , its amplitude can be written as @xmath79 \hspace*{3.6cm}\nonumber\\ -f_{\pi}v_{tb}^{*}v_{td}f_{e}^{p}\left[-\frac{1}{3}c_{5}-c_{6}+\frac{1}{6}c_{7 } + \frac{1}{2}c_{8}\right ] + m_{e}\left[v_{ud}v_{ub}^{*}c_{2}-v_{tb}^{*}v_{td}(-c_{3}+\frac{1}{2}c_{9 } ) \right]\hspace*{1cm}\nonumber\\ -v_{tb}^{*}v_{td}m_{e}^{p}\left(\frac{1}{2}c_{7}-c_{5}\right)-v_{tb}^{*}v_{td}m_{a } \left(\frac{1}{2}c_{9}-c_{3}\right ) -v_{tb}^{*}v_{td}m_{a}^{p}\left(\frac{1}{2}c_{7}-c_{5}\right)\hspace*{2 cm } \nonumber\\ -f_{b}v_{tb}^{*}v_{td}f_{a}\left[-\frac{1}{3}c_{3}-c_{4}+\frac{1}{6}c_{9 } + \frac{1}{2}c_{10}\right ] -f_{b}v_{tb}^{*}v_{td}f_{a}^{p}\left[-\frac{1}{3}c_{5}-c_{6 } + \frac{1}{6}c_{7}+\frac{1}{2}c_{8}\right].\end{aligned}\ ] ] and the decay width is then expressed as @xmath80
the following parameters have been used in our numerical calculation @xcite : @xmath81 we leave the ckm phase angle @xmath82 as a free parameter , whose definition is @xmath83.\ ] ] in this language , the decay amplitude of @xmath84 in eq.([eq : width ] ) can be parameterized as @xmath85,\label{a1}\ ] ] where @xmath86 , and @xmath57 is the relative strong phase between tree diagrams @xmath87 and penguin diagrams @xmath88 . @xmath89 and @xmath57 can be calculated from pqcd . using the above parameters in ( [ eq : shapewv ] ) , we get @xmath90 and @xmath91 from pqcd calculation , which shows the dominance of the tree contribution in this decay and a large strong phase calculated from pqcd . similarly , the decay amplitude for @xmath92 can be parameterized as @xmath93.\label{a2}\ ] ] therefore the averaged decay width for @xmath94 is @xmath95 .
\label{eq : width3}\end{aligned}\ ] ] it is a function of @xmath96 . in fig .
[ figure : fig2 ] , we plot the averaged branching ratio of the decay @xmath97 with respect to the parameter @xmath98 .
since the latest experiment constraint upon the ckm angle @xmath98 from belle and babar is @xmath98 around @xmath99 @xcite , we can arrive from fig .
[ figure : fig2 ] : @xmath100 previous naive and generalized factorization approach gives a similar branching ratios at @xmath101 with the form factor @xmath102 @xcite . in paper @xcite , beneke _
_ also calculate this decay mode using qcd improved factorization approach ( bbns ) .
it is based on naive factorization approach .
the dominant contribution is still proportional to @xmath15 form factor , which is introduced as an input parameter . in principal , the decay amplitude expand as series of @xmath103 and @xmath104 .
but in practice , only the first order of @xmath103 corrections is calculated , including the so called non - factorizable contributions .
the annihilation type contribution is power ( @xmath104 ) suppressed in bbns approach .
therefore , the branching ratio predicted in qcd factorization and pqcd should not differ too much ; but the cp violation in these two approaches will be different , since it depends on many non - leading order contributions ( see below for discussion ) . in ref.@xcite ,
the branching ratio is about @xmath105 , which is larger than our pqcd result and previous fa method @xcite , because their form factor @xmath106 @xcite is larger than the previous factorization approach and our calculation below . the diagrams ( a ) and ( b ) in fig .
[ figure : fig1 ] correspond to the @xmath107 transition form factor @xmath108 , where @xmath109 is the momentum transfer .
the sum of their amplitudes have been given by eq .
( [ fe ] ) , so we can use pqcd approach to compute this form factor . our result is @xmath110 , if @xmath111 ; and @xmath112 , if @xmath113 . in our approach , this form factor is sensitive to the decay constant and wave function of @xmath6 meson , where there is large uncertainty ; but not sensitive to the @xmath33 meson wave function .
eventually this form factor can be extracted from semi - leptonic experiments @xmath114 in the future . in our calculation , the only input parameters are wave functions , which stand for the non - perturbative contributions . up to now , no exact solution is made for them .
so the main uncertainty in pqcd approach comes from @xmath115 wave functions . in this paper
, we choose the light cone wave functions which are obtained from qcd sum rules @xcite . for @xmath34 meson
, the distribution amplitude of light cone wave function should take asymptotic form if the energy scale @xmath116 .
but in our case , the scale is not more than @xmath117gev , so we choose the corrected asymptotic form for twist 2 distribution amplitude @xmath118 , and other twist 3 distribution amplitudes derived using equation of motion by neglecting three particle wave functions @xcite .
these functions are listed in the appendix , which are also used in decay mode @xmath119 @xcite and @xmath120 @xcite etc .
we also try to use the asymptotic form for @xmath34 meson , for all the three distribution amplitudes @xmath118 , @xmath121 and @xmath122 , since we have very poor knowledge about twist 3 distribution amplitudes @xcite .
the branching ratio of @xmath123 is nearly unchanged ( only @xmath124 ) , because the branching ratio of @xmath125 is mainly determined by the form factor @xmath126 ( see fig.1(a ) and ( b ) ) which is not dependent on @xmath34 wave function .
however , the cp asymmetry changes from @xmath127 to @xmath128 by @xmath129 , when @xmath130 .
this is because the direct cp asymmetry depend on the strong phase ( see discussion below ) , which comes from non - factorizable and annihilation diagrams , where all three meson wave functions are involved .
the cp asymmetry predicted here should be used with great care , since it depends on two much uncertainties . for heavy @xmath9 and @xmath6 meson ,
its wave function is still under discussion using different approaches @xcite . in this paper
, we find the branching ratio of @xmath131 is sensitive to the wave function parameter @xmath132 .
for @xmath133 , the resulted branching ratio will decrease from about @xmath134 to about @xmath135 .
when we set @xmath113 , our result is more closer to that of qcd factorization @xcite .
this sensitive dependence should be fixed by the @xmath15 form factors from the semi - leptonic @xmath6 decays .
other uncertainties in our calculation include the next - to - leading order @xmath103 qcd corrections and higher twist contributions , which need more complicated calculations . from our calculation , we find that the dominant contribution comes from tree level diagrams ( see fig.1 ( a ) and ( b ) ) in this decay .
if su(3 ) symmetry is good , the branching ratio of @xmath136 should be equal to that of @xmath137 .
the experimental result of @xmath137 is @xmath138 @xcite
. the predicted branching ratio of @xmath16 is about 1.7 times that of @xmath139 , where the difference comes mainly from su(3 ) symmetry breaking : the decay constant @xmath140 larger than @xmath141 and @xmath142 larger than @xmath143 . in the calculation , we also find that the electroweak - penguins contribution is negligibly small as @xmath144 in branching ratio . for the experimental side ,
there is recent upper limit on the decay @xmath145 @xcite , @xmath146 at 90% c.l .
our predicted result is consistent with this upper limit .
for the decays of @xmath147 , the tree level contribution is suppressed due to the small wilson coefficients @xmath148 .
thus the penguin diagram contribution is comparable with the tree contribution .
we study the averaged branching ratio of the decay @xmath149 as a function of @xmath98 in fig .
[ figure : fig3 ] .
it is similar with fig.[figure : fig2 ] .
we find that the branching ratio of @xmath149 is about @xmath150 when @xmath98 is near @xmath99 , it is a little smaller than the result of ref . @xcite . in sm ,
the ckm phase angle is the origin of cp violation . using eqs.([a1 ] ) and
( [ a2 ] ) , the direct cp violation parameter can be derived as @xmath151 it is approximately proportional to ckm angle @xmath152 , strong phase @xmath153 and the relative size @xmath89 between penguin contribution and tree contribution .
we show the direct cp violation parameters as a function of ckm angle @xmath98 in fig .
[ figure : fig4 ] . from this figure
one can see that the direct cp asymmetry parameter of @xmath154 and @xmath155 can be as large as @xmath156 and @xmath157 when @xmath98 is near @xmath158 .
the larger direct cp asymmetry of @xmath159 decay is mainly due to a larger @xmath89 in @xmath159 than in @xmath154 .
the direct cp asymmetry predicted in qcd factorization approach is quite different from our result , due to the different source of strong phases . in qcd factorization approach ,
the strong phase mainly comes from the perturbative charm quark loop diagram , which is @xmath103 suppressed @xcite . while the strong phase in pqcd comes mainly from non - factorizable and annihilation type diagrams .
the sign of the direct cp asymmetry is different for these two approaches in @xmath160 decay , and the magnitude of cp asymmetry in qcd factorization ( about 5% ) is also smaller than pqcd .
the future lhc - b experiments can make a test for the two methods . for the decays of @xmath147 ,
the final @xmath161 mesons can not be detected directly .
what the experiments measured are their mixtures @xmath162 and @xmath163 , thus a mixing induced cp violation is involved . following notations in the previous literature @xcite , we define the mixing induced cp violation parameter as @xmath164 where @xmath165 using unitarity condition of the ckm matrix @xmath166 , and eqs.([a1],[a2 ] ) , we can get @xmath167 where @xmath168 . combining eq.([lam ] ) and ( [ mcp ] ) , we can get @xmath169 if @xmath170 is a very small number , the mixing induced cp asymmetry is proportional to @xmath171 , which will be a good place for the ckm angle @xmath172 measurement .
however as we already mentioned , the tree contribution in this channel is suppressed , @xmath173 is a large number , so that the @xmath174 behavior is dominant in the eq .
( [ cpf ] ) .
the result of mixing induced cp violation is shown in fig .
[ cp ] , which is indeed a roughly @xmath174 behavior .
the tail near @xmath175 also shows the contribution from @xmath171 in eq.([cpf ] ) .
in this work , we study the branching ratio and cp asymmetry of the decays @xmath97 and @xmath149 in pqcd approach . from our calculation , we find that the branching ratio of @xmath97 is about @xmath176 ; @xmath4 around @xmath177 and there are large cp violation in the processes , which may be measured in the future lhc - b experiments and btev experiments at fermilab .
the authors thank m - z yang for helpful discussions , they also thank professor dong - sheng du for reading the manuscript .
this work is partly supported by national science foundation of china under grant no .
90103013 , 10475085 and 10135060 .
[ appendix ]
in the appendix we present the explicit expressions of the formulas used in section ii . first , we give the expressions of the meson distribution amplitudes @xmath50 . for @xmath178 meson wave function ,
we use the similar wave function as @xmath9 meson @xcite : @xmath179.\label{waveb}\ ] ] we set the central value of parameter @xmath180 in our numerical calculation , and @xmath181 is the normalization constant using @xmath182 .
we use the distribution amplitude @xmath186 of the k meson from ref .
@xcite : @xmath187 , \\
\nonumber \phi_{k}^p(x ) & = & \frac{f_{k}}{2\sqrt{2 n_c } } [ 1 + 0.106(3t^2 - 1)-0.148(3 - 30t^2 + 35t^4)/8 ] , \\
\phi_{k}^t(x ) & = & \frac{f_{k}}{2\sqrt{2 n_c}}t[1 + 0.1581(5t^2 - 3)],\end{aligned}\ ] ] whose coefficients correspond to @xmath188 . in our numerical analysis , we use the one loop expression for the strong running coupling constant , @xmath189 where @xmath190 and @xmath191 is the number of active quark flavor at the appropriate scale @xmath19 .
@xmath192 is the qcd scale , which we take @xmath193mev at @xmath194 .
@xmath195 , @xmath196 , @xmath197 used in the decay amplitudes are defined as @xmath198 where the so called sudakov factor @xmath199 resulting from the resummation of double logarithms is given as @xcite @xmath200 \label{su1}\ ] ] with @xmath201 \left(\frac{\alpha_s}{\pi}\right)^2 , \\
b=\frac{2}{3}\frac{\alpha_s}{\pi}\ln\left(\frac{e^{2\gamma_{e}-1}}{2}\right).\hspace{6cm}\end{gathered}\ ] ] here @xmath202 is the euler constant , @xmath191 is the active quark flavor number . for the detailed derivation of the sudakov factors ,
@xcite .
the functions @xmath203 come from the fourier transformation of propagators of virtual quark and gluon in the hard part calculations .
they are given as @xmath204 , \label{eq : propagator1}\end{aligned}\ ] ] we adopt the parametrization for @xmath213 contributing to the factorizable diagrams @xcite , @xmath214^{c } , \hspace{0.5cm}c=0.3.\end{aligned}\ ] ] the hard scale @xmath215 in eq.([fe])-([fap ] ) are chosen as @xmath216 they are given as the maximum energy scale appearing in each diagram to kill the large logarithmic radiative corrections .
m. wirbel , b , stech , and m. bauer , z. phys .
c29 , 637 ( 1985 ) ; + m. bauer , b , stech , and m. wirbel , _
ibid._34 , 103 ( 1987 ) ; + l .- l .
chau , h .- y .
cheng , w.k .
sze , h. yao , and b. tseng , phys .
d43 , 2176 ( 1991 ) ; 58 , 019902(e ) ( 1998 ) .
a. ali , g. kramer and c.d .
l , phys . rev .
d 58 , 094009 ( 1998 ) ; + _ ibid . _ 59 , 014005 ( 1999 ) ; c.d .
l , nucl . phys .
b ( proc . suppl . ) 74 , 227 ( 1999 ) . y .- h . chen , h .- y .
cheng , b. tseng , and k .- c .
yang , phys .
d60 , 094014 ( 1999 ) ; + h .- y .
cheng and k .- c .
yang , _ ibid .
_ 62 , 054029 ( 2000 ) .
m. beneke , g. buchalla , m. neubert , and c.t .
sachrajda , phys .
83 , 1914 ( 1999 ) ; nucl .
b591 , 313 ( 2000 ) . a.j .
schwartz(for the belle collaboration ) , hep - ex/0411075 ; + a. bevan ( for the babar collaboration ) , hep - ex/0411090 .
xing , phys . rev .
d48 , 3400 ( 1993 ) ; d.s .
yang , phys .
b358 , 123 ( 1995 ) ; y.h .
chen , h.y , cheng , b. tseng , phys .
d59 , 074003 ( 1999 ) .
m. beneke and m. neubert , nucl .
b 675 , 333 ( 2003 ) ; + j .- f.sun , g.h . zhu , d .- s .
du , phys .
d68 , 054003 ( 2003 ) .
a. khodjamirian , t. mannel and m. melcher , phys .
d70 , 094002 ( 2004 ) ; v.m .
braun and a. lenz , phys .
d70 , 074020 ( 2004 ) .
t. huang , x .- h .
zhou , phys .
d70 , 014013 ( 2004 ) ; + t. huang , x .-
wu , x .- h .
wu , phys .
d70 , 053007 ( 2004 ) ; + t. huang , x .-
wu , phys .
d70 , 093013 ( 2004 ) ; + t. huang , m .- z .
zhou , x .- h .
wu , hep - ph/0501032 .
h.kawamura , j.kodaira , c - f qiao and k.tanaka , nucl.phys.proc.suppl . 116 269(2003 ) ; + h .-
liao , phys .
d70 , 074030(2004 ) ; + tao huang , xing - gang wu and ming - zhen zhou , phys .
b611 , 260(2005 ) ; + bodo geyer and oliver witzel , hep - ph/0502239 .
cleo collaboration , d. cronin - hennessy et al.,hep - ex/0001010 .
li and k. ukai , phys .
b555 , 197 ( 2003 ) .
n . li and b. melic , eur .
j. c11 , 695 ( 1999 ) .
n . li , phys .
d52 , 3958 ( 1995 ) .
t. kurimoto , h .-
li , and a.i .
sanda , phys .
d65 , 014007 ( 2002 ) . | in the framework of perturbative qcd approach , we calculate the branching ratio and cp asymmetry for @xmath0 and @xmath1 decays . besides the usual factorizable diagrams ,
both non - factorizable and annihilation type contributions are taken into account .
we find that ( a ) the branching ratio of @xmath2 is about @xmath3 ; @xmath4 about @xmath5 ; and ( b ) there are large cp asymmetries in the two processes , which can be tested in the near future lhc - b experiments at cern and btev experiments at fermilab . |
the teichmller space @xmath3 of the unit disk @xmath4 , called the _ universal _ teichm " uller space , consists of all quasisymmetric maps @xmath5 which fix @xmath6 , @xmath7 and @xmath8 on the unit circle @xmath9 ( cf .
the teichm " uller space of an arbitrary hyperbolic surface embeds in @xmath3 as a complex banach submanifold .
thurston s boundary to the universal teichmller space @xmath3 is the space of projective bounded measured laminations @xmath10 of @xmath4 ( cf .
@xcite , @xcite ) .
we study the limits of teichmller geodesic rays on thurston s boundary to @xmath3 .
bonahon @xcite defined an embedding of the teichmller space @xmath11 of a closed surface @xmath12 of genus at least two into the space of geodesic currents ( equipped with the weak * topology ) .
the space of asymptotic rays to the image of the embedding of @xmath11 is identified with the space of projective measured lamination of @xmath12-thurston s boundary to @xmath11 .
the universal teichm " uller space @xmath3 embeds into the space of geodesic currents of @xmath4 when geodesic currents are equipped with the _ uniform _ weak * topology ( cf .
@xcite , @xcite , @xcite ) and this embedding is real analytic ( cf
. otal @xcite ) .
the image of @xmath3 in the space of geodesic currents is closed and unbounded , and the space of asymptotic rays to the image of @xmath3-thurston s boundary to @xmath3- is identified with the projective bounded measured laminations @xmath10 ( cf .
@xcite , @xcite ) .
in particular , the earthquake paths @xmath13 as @xmath14 accumulate to their corresponding projective earthquake measures @xmath15\in pml_{bdd}(\mathbb{d})$ ] in the uniform weak * topology ( cf .
@xcite , @xcite ) .
the construction of thurston s boundary works for all hyperbolic surfaces simultaneously since any invariance under a fuchsian group is preserved under the construction . in the case of closed surfaces ,
masur @xcite proved that the teichmller geodesic rays obtained by shrinking the vertical trajectories of holomorphic quadratic differentials with uniquely ergodic vertical foliations converge to the projective classes of their vertical foliations in thurston s boundary . on the other hand ,
if the vertical foliation of a holomorphic quadratic differential consists of finitely many cylinders then the limit of the teichmller geodesic on thurston s boundary is the projective class of the measured lamination supported on finitely many simple closed geodesics homotopic to the cylinders with equal weights ( cf .
@xcite ) . in both cases
the teichmller geodesic rays have unique endpoints on thurston s boundary and the endpoints depend only on the vertical foliations .
however , when the vertical foliations of holomorphic quadratic differentials on closed surfaces are not uniquely ergodic then the limit sets of the corresponding teichmller rays consist of more than one point ( cf .
lenzhen @xcite , leininger - lenzhen - rafi @xcite , and chaika - masur - wolf @xcite ) .
we consider the limits of teichmller geodesic rays in the universal teichmller space @xmath3 corresponding to integrable holomorphic quadratic differentials . in our previous work we showed that when a holomorphic quadratic differential has no zeroes in @xmath4 , and the natural parameter maps the unit disk onto a domain in @xmath16 between the graphs of two functions , then the teichmller geodesic ray has a unique endpoint on thurston s boundary of @xmath3 , but the endpoint depends on both vertical and horizontal foliations of @xmath17 ( cf .
we extend this result to all integrable holomorphic quadratic differentials on the unit disk @xmath4 .
the hyperbolic plane is identified with the unit disk @xmath4 and the visual boundary of the hyperbolic plane is identified with the unit circle @xmath9 .
a ( hyperbolic ) geodesic in @xmath4 is uniquely determined by it endpoints ; the space of geodesics of @xmath4 is identified with @xmath18 .
let @xmath17 be an arbitrary integrable holomorphic quadratic differential on the unit disk @xmath4 .
each vertical trajectory of @xmath17 has two distinct endpoints on the boundary circle @xmath9 of the unit disk @xmath4 ( cf .
thus each vertical trajectory of @xmath17 is homotopic to a unique geodesic of @xmath4 relative ideal endpoints on @xmath9 .
let @xmath19 be the set of the geodesics in @xmath4 homotopic to the vertical trajectories of @xmath17 ( cf .
figure 1 ) . given
a box of geodesics @xmath20\times [ c , d]\subset \mathbb{s}^1\times \mathbb{s}^1-diag$ ] , denote by @xmath21\times [ c , d]}$ ] any ( at most countable ) union of sub - arcs of horizontal trajectories that intersects exactly once each vertical trajectory of @xmath17 with one endpoint in @xmath20 $ ] and the other endpoint in @xmath22 $ ] , and that does not intersect any other vertical trajectories of @xmath17 .
we define a measured lamination @xmath23 of @xmath4 supported on @xmath19 by @xmath24\times [ c , d])=\int_{i_{[a , b]\times [ c , d]}}\frac{1}{l(x)}dx,\end{aligned}\ ] ] where @xmath25 is the @xmath17-length ( i.e. the length induced by @xmath26 ) of a vertical trajectory through @xmath27\times [ c , d]}$ ] and the integration is in the natural parameter of @xmath17 .
we obtain ( cf .
proposition [ prop : common_endpoints ] and proof of theorem [ thm : weak*convergence ] ) .2 cm * proposition 1 . *
_ let @xmath23 be the measured lamination homotopic to the vertical foliation of an integrable holomorphic quadratic differential @xmath17 on @xmath4 defined by the above integration .
then @xmath28\times [ c , d]}\mu_{\varphi}([a , b]\times [ c , d])<\infty\ ] ] where the supremum is over all boxes of geodesics @xmath20\times [ c , d]\subset \mathbb{s}^1\times \mathbb{s}^1-diag$ ] with @xmath29 . _
the measured lamination @xmath23 satisfies @xmath30)=0\ ] ] for all @xmath31 and @xmath22\subset \mathbb{s}^1 $ ] , and in particular @xmath23 does not have atoms .
.2 cm for @xmath32 , let @xmath33 be the teichmller mapping that shrinks the vertical trajectories of @xmath17 by a multiplicative constant @xmath34 .
the _ teichmller geodesic ray _ @xmath35 as @xmath36 leaves every bounded subset of the universal teichmller space @xmath3 .
we obtain ( cf .
theorem [ thm : weak*convergence ] ) .2 cm * theorem 1 .
* _ let @xmath37 be the teichmller geodesic ray in @xmath3 that shrinks the vertical trajectories of an integrable holomorphic quadratic differential @xmath17 by a multiplicative constant @xmath32 .
then @xmath38\in pml_{bdd}(\mathbb{d})\ ] ] as @xmath39 in thurston s closure @xmath40 of @xmath3 , where @xmath23 is the measured lamination defined by equation ( [ def : lamination ] ) and the convergence is in the weak * topology on geodesic currents .
_ in particular , the limit set of any teichmller ray in @xmath3 consists of a unique point .
.2 cm * remark 1 .
* the limit point @xmath23 depends on the vertical foliation and on the lengths of the vertical trajectories unlike for closed surfaces .
the lengths of vertical trajectories are given by the transverse measure to the horizontal foliation .
therefore the limit point depends on both vertical and horizontal foliations of @xmath17 which is a new phenomenon that does not appear for closed surfaces .
.2 cm * remark 2 .
* the measure @xmath41\times [ c , d])$ ] is in fact the conformal modulus of the family of vertical trajectories of @xmath17 with one endpoint in @xmath20 $ ] and another endpoint in @xmath22 $ ] ( cf .
proposition [ prop : vertical_mod_measure ] ) .
.2 cm * remark 3 .
* the above theorem is motivated by the results of masur @xcite in the case of a closed surface .
the major difference in this work is that the hyperbolic plane has no closed geodesics and that the universal teichmller space is infinite dimensional non - separable banach manifold .
the convergence questions that arise in this setup and the methods applied are of a more analytic nature than for closed surfaces .
moreover , we emphasise the existence of a unique limit point for any teichmller ray in @xmath3 which is not true for the teichmller spaces of closed surfaces .
.2 cm the convergence of teichmller geodesic rays is in the weak * topology while the convergence of earthquake paths is in the _ uniform _ weak * topology .
we prove in [ sec : ce ] .2 cm * proposition 2 . * _ there exists an integrable holomorphic quadratic differential @xmath17 on the unit disk @xmath4 such that the corresponding teichmller geodesic ray does not converge in the _ uniform _ weak * topology . _
.2 cm * remark 4 .
* thus while any teichmller geodesic ray in @xmath3 converges in the weak * topology , there exist teichmller rays that do not converge in the strongest possible sense ( the uniform weak * topology ) in @xmath40 unlike earthquake paths .
also note that the uniform weak * topology and the weak * topology agree on the space of geodesic currents of a closed surface .
.2 cm denote by @xmath42 the space of all integrable holomorpic quadratic differentials on the unit disk @xmath4 .
let @xmath43 , where @xmath44 if there exists @xmath45 with @xmath46 . by definition
, we have @xmath47 for any @xmath45 .
therefore we obtained a map @xmath48 given by @xmath49)=[\mu_{\varphi}],\ ] ] where @xmath50 $ ] and @xmath51 $ ] are the projective classes of @xmath17 and @xmath23 , respectively .
.2 cm * theorem 2 . *
_ the map @xmath52\mapsto [ \mu_{\varphi}]\ ] ] is injective . _ .2 cm
* remark 5 .
* by theorem 1 and theorem 2 , two different teichmller geodesic rays in @xmath3 starting at the basepoint of @xmath3 converge to different points in thurston s boundary . on the other hand ,
masur @xcite proved that two teichmller geodesic rays corresponding to two holomorphic quadratic differentials whose vertical foliations decompose a compact surface into finitely many cylinders of the same topological type but different relative heights converge to the same point in thurston s boundary .
.2 cm for @xmath53 , denote by @xmath54 the measured lamination whose support is the geodesic lamination @xmath19 homotopic to the vertical foliation of @xmath17 and whose transverse measure is induced by the transverse measure to the vertical foliation induced by @xmath17 .
we recover the integral of @xmath55 using @xmath23 and @xmath54 , namely @xmath56 .2 cm * theorem 3 . * _ the map @xmath57 defined by @xmath58 is injective .
_ .2 cm the paper is organized as follows . in
2 we define the universal teichmller space @xmath3 , the space of geodesic currents , the liouville current and thurston s boundary to @xmath3 . in
3 we define modulus of a family of curves and find a relationship between the modulus and relative distance .
finally we give asymptotic relationship between the modulus and the liouville current which is fundamental to our work . in
4 we study the limits of teichmller geodesic rays and prove theorem 1 . in
5 we give a counter - example to uniform weak * convergence of teichmller geodesic rays . in 6
we study the relationship between the integrable holomorphic quadratic differentials and two measured laminations homotopic to the vertical foliation .
.2 cm _ acknowledgements .
_ we would like to thank mr .
huiping pan for his useful comments .
we identify the unit disk @xmath4 with the hyperbolic plane ; the visual boundary to @xmath4 is the unit circle @xmath9 . a homeomorphism @xmath5 is said to be _ quasisymmetric _ if there exists @xmath59 such that @xmath60 for all circular arcs @xmath61 with a common boundary point and disjoint interiors such that @xmath62 , where @xmath63 is the length of @xmath64 .
a homeomorphism is quasisymmetric if and only if it extends to a quasiconformal map of the unit disk , see e.g. @xcite .
the universal teichmller space @xmath3 consists of all quasisymmetric maps @xmath5 that fix @xmath65 . if @xmath66 is a quasiconformal map , denote by @xmath67 its quasiconformal constant .
the teichmller metric on @xmath3 is given by @xmath68 , where @xmath69 runs over all quasiconformal extensions of the quasisymmetric map @xmath70 .
the teichmller topology is induced by the teichmller metric .
bonahon s approach @xcite to thurston s boundary of the teichmller space @xmath11 of a closed surface @xmath12 is to embed @xmath11 into the space of geodesic currents on @xmath12 .
geodesic current _ on @xmath12 is a positive borel measure on the space of geodesics @xmath71 of the universal covering @xmath4 of @xmath12 that is invariant under the action of the covering group @xmath72 .
each point in the teichmller space @xmath11 is a quasisymmetric map @xmath73 that conjugates the covering fuchsian group @xmath72 onto another fuchsian group .
the _ liouville measure _
@xmath74 on the space of geodesic of @xmath4 is given by @xmath75 for any borel set @xmath76 . if @xmath77\times [ c , d]$ ] is a _ box of geodesics
_ then @xmath78\times [ c , d])=\log\frac{(a - c)(b - d)}{(a - d)(b - c)}.\ ] ] to each quasisymmetric map @xmath5 that conjugates @xmath72 onto another fuchsian group , we assign the pull back @xmath79 of the liouville measure .
this assignment is a homeomorphism of @xmath11 onto its image in the space of geodesic currents for @xmath12 when equipped with the weak * topology ( cf .
the set of the asymptotic rays to the image of @xmath11 is identified with the projective measured laminations on @xmath12-the thurston s boundary of @xmath11 ( cf .
@xcite ) .
the universal teichmller space @xmath3 maps into the space of geodesic currents by taking the pull backs by quasisymmetric maps of the liouville measure .
there is no invariance condition on the quasisymmetric maps or on the pull backs of the liouville measure .
a geodesic current @xmath80 is _ bounded _ if @xmath81\times [ c , d]}\alpha ( [ a , b]\times [ c , d])<\infty\ ] ] where the supremum is over all boxes of geodesics @xmath20\times [ c , d]$ ] with @xmath82\times [ c , d])=\log 2 $ ]
. the pull backs @xmath79 for @xmath83 quasisymmetric are bounded geodesic currents .
the space of bounded geodesic currents is endowed with the family of hlder norms parametrized with the hlder exponents @xmath84 ( cf .
the pull backs of the liouville measure define a homeomorphism of @xmath3 onto its image in the bounded geodesic currents ; the homeomorphism is differentiable with a bounded derivative ( cf .
@xcite ) and , in fact , otal @xcite proved that the embedding is real - analytic .
the asymptotic rays to the image of @xmath3 are identified with the space of projective bounded measured laminations ( cf .
thus thurston s boundary of @xmath3 is the space @xmath10 of all projective bounded measured laminations on @xmath4 ( and an analogous statement holds for any hyperbolic riemann surface ) .
alternatively , the space of geodesic currents can be endowed with the uniform weak * topology ( for definition cf .
@xcite ) and thurston s boundary for @xmath3 is again @xmath10 ( cf .
let @xmath85 be a family of rectfiable curves in @xmath16 . an _
admissible metric _
@xmath86 for @xmath87 is a non - negative borel measurable function on @xmath4 such that the @xmath86-length of each @xmath88 is at least one , namely @xmath89 the _ modulus _ @xmath90 of the family @xmath87 is given by @xmath91 where the infimum is over all admissible metrics @xmath86
. we will mostly be interested in estimating moduli of families of curves in a domain @xmath92 connecting two subsets of the boundary of @xmath93 .
thus , given @xmath94 we denote @xmath95 the family of rectifiable curves @xmath96 having one endpoint in @xmath97 and the other endpoint in @xmath98 . when @xmath93 is the unid disc @xmath4 and @xmath99 is a quadruple of distinct points on the boundary circle @xmath9 given in the counterclockwise order we denote @xmath100\times[c , d ] } = ( ( a , b),(c , d);\mathbb{d}).\ ] ] lemma [ lemma : modproperties ] below , summarizes some of the main properties of the modulus , which we will use repeatedly throughout the paper .
we refer the reader to @xcite for the proofs of these properties below and for further background on modulus .
if @xmath101 and @xmath102 are curve families in @xmath16 , we will say that @xmath101 _ overflows _ @xmath102 and will write @xmath103 if every curve @xmath104 contains some curve @xmath105 .
[ lemma : modproperties ] let @xmath106 be curve families in @xmath16 .
then * monotonicity : if @xmath107 then @xmath108 . *
subadditivity : @xmath109 * overflowing : if @xmath110 then @xmath111 .
heuristically modulus of @xmath95 measures the amount of curves connecting @xmath97 and @xmath98 in the @xmath93 .
the more short " curves there are the bigger the modulus is .
this heuristic may be made precise using a notion of relative distance @xmath112 , which we define next .
given two continua @xmath97 and @xmath98 in @xmath16 we denote @xmath113 i.e. @xmath112 is the _ relative distance _ between @xmath97 and @xmath98 in @xmath16 .
[ lemma : mod - reldist ] for every pair of continua @xmath114 we have @xmath115 let @xmath116 and @xmath117 be the family of curves @xmath118 such that @xmath119 and @xmath120 .
then @xmath121 and similarly @xmath122 .
therefore , @xmath123 denoting by @xmath124 the @xmath125-neighborhood of the set @xmath126 , we note , that @xmath127 is admissible for @xmath128 .
therefore , we have @xmath129 where @xmath130 is the euclidean area of @xmath124 . combining inequalities ( [ modest : min ] ) and ( [ modest : reldist1 ] ) we obtain ( [ modest : reldistance ] ) .
let @xmath131 and @xmath132 , @xmath133 be a sequence of pairs of continua in @xmath16 . if the sequence @xmath134 is bounded away from @xmath135 then @xmath136 is bounded .
the previous lemma is very weak for large @xmath112 , since it is in fact easy to see that @xmath137 tends to @xmath135 as @xmath138 .
but we will not need this estimate in the present paper and will refer the interested reader to heinonen s book @xcite for relations between the modulus and relative distance .
the following lemma is an easy consequence of the asymptotic properties of the moduli ( cf .
@xcite ) .
[ lem : mod_liouville_measure ] let @xmath99 be a quadruple of points on @xmath9 in the counterclockwise order .
let @xmath139\times [ c , d]}$ ] consist of all differentiable curves @xmath140 in @xmath4 which connect @xmath20\subset \mathbb{s}^1 $ ] with @xmath22\subset \mathbb{s}^1 $ ]
. then @xmath141\times [ c , d]})-\frac{1}{\pi}\mathcal{l}([a , b]\times [ c , d])-\frac{2}{\pi}\log 4\to 0\ ] ] as @xmath142\times [ c , d]})\to\infty$ ] , where @xmath74 is the liouville measure .
note that simultaneously @xmath142\times [ c , d]})\to\infty$ ] and @xmath82\times [ c , d])\to\infty$ ] .
let @xmath17 be an integrable holomorphic quadratic differential on the unit disk @xmath4 .
in other words , @xmath143 is holomorphic and @xmath144 a point @xmath145 is said to be _ regular _ for @xmath146 if @xmath147 . in a neighborhood of every regular point of @xmath17
the parameter @xmath148 given by path integral @xmath149 is called a _ natural parameter _ for @xmath17 .
the holomorphic quadratic differential @xmath150 has representation @xmath151 in the natural parameter @xmath148 .
moreover , if @xmath152 is another natural parameter then @xmath153 at their intersection ( cf .
@xcite ) . a _ vertical arc _ for @xmath17 is a differentiable arc @xmath154 that passes only through regular points of @xmath17 and that satisfies @xmath155 for all @xmath156 .
equivalently , a vertical arc is an inverse image of a euclidean vertical arc in the natural parameter @xmath148 .
a _ vertical trajectory _ of @xmath17 is a _
maximal _ vertical arc .
similarly , a _ horizontal arc _ for @xmath17 is a differentiable arc @xmath154 that passes through regular points and satisfies @xmath157 for all @xmath156 ( cf .
@xcite ) .
each end of a vertical trajectory either accumulates to a zero of @xmath158 or to the boundary @xmath9 of @xmath4 .
in particular , if an end of a vertical trajectory of @xmath17 accumulates to the boundary @xmath9 then the limit set on @xmath9 consists of a single point and we say that the vertical trajectory has an endpoint on @xmath9 ( cf .
the set of zeroes of @xmath17 is countable and therefore only countably many vertical trajectories have an endpoint at a zero of @xmath17 .
any vertical trajectory of @xmath17 not in the above countable set has two distinct endpoints on @xmath9 ( cf .
@xcite ) .
we define the _ width _ of a curve @xmath140 in @xmath4 . by strebel @xcite
, the unit disk @xmath4 can be decomposed into countably many disjoint open strips @xmath159 up to a countable family of vertical trajectories , where @xmath160 is an open horizontal arc and @xmath161 is the union of vertical trajectories intersecting @xmath160 .
the strips @xmath159 are open and simply connected .
the natural parameter @xmath162 is well - defined on each @xmath159 since @xmath159 is simply connected and does not contain any zeroes of @xmath17 .
any borel @xmath163 has well - defined width @xmath164 if @xmath165 , denote by @xmath166 the projection of @xmath140 onto @xmath160 along the vertical trajectories .
then the width of @xmath140 is defined by @xmath167 assume that @xmath140 is not contained in a single strip . consider the collection of borel sets @xmath168 for all @xmath7 with @xmath169 .
we define the _ width _ of @xmath140 by @xmath170 the definition @xmath171 is given in terms of the strips @xmath159 . to see that @xmath171 is independent of the choice of the strips , let @xmath172 be another countable collection of disjoint open strips that covers @xmath4 up to countable union of singular vertical trajectories .
the two strips @xmath159 and @xmath172 are either disjoint or they intersect in an open strip @xmath173 , where @xmath174 is an open subinterval on @xmath160 which can be homotoped modulo vertical trajectories to subinterval of @xmath175 .
the homotopy is measure preserving for @xmath176 .
since @xmath177 is at most countable ( which is of measure zero ) , it follows that @xmath178 this implies that @xmath171 is independent of the choice of the covering by the strips .
[ prop : upper_bound ] let @xmath179\times [ c , d])$ ] be the family of rectifiable arcs in @xmath4 with one endpoint in @xmath20\subset \mathbb{s}^1 $ ] and the other endpoint in @xmath22\subset \mathbb{s}^1 $ ]
. denote by @xmath33 the teichmller map of @xmath4 that shrinks the vertical trajectories of @xmath17 by the multiplicative constant @xmath180 . then @xmath181,[c , d]))\ ] ] where @xmath182,[c , d])$ ] is the set of vertical trajectories of @xmath17 with one endpoint in @xmath20 $ ] and the other endpoint in @xmath22 $ ] . by strebel @xcite , almost every point of @xmath9
is at a finite distance from an interior point of @xmath4 in the path metric @xmath183 , called @xmath17_-metric_. let @xmath184 be on finite distances from an interior point such that @xmath20\subset [ a',b']$ ] and @xmath22\subset [ c',d']$ ] .
let @xmath185\times [ c',d'])$ ] . ,
@xmath186 and @xmath187.,width=226 ] namely , let ( cf . figure 2 ) @xmath188\mbox { and } [ c',d']\}.\ ] ] since @xmath189 , we have @xmath190 .
let @xmath191 and @xmath192 be two simple non - intersecting differentiable arcs in @xmath4 with endpoints @xmath193 and @xmath194 , respectively .
let @xmath195 be the subset of @xmath4 with boundary consisting of arcs @xmath191 , @xmath196\subset \mathbb{s}^1 $ ] , @xmath192 and @xmath197\subset \mathbb{s}^1 $ ] .
let @xmath198 be the family of rectifiable curves in @xmath195 that connect @xmath191 and @xmath192 .
then the family @xmath186 overflows the family @xmath187 and we have @xmath199 fix @xmath200 and define @xmath201 and @xmath202 by the subadditivity of the modulus @xmath203 first consider @xmath204 .
define the metric @xmath205 for @xmath206 , where @xmath207 is the terminal holomorphic quadratic differential on @xmath208 ( cf .
recall that the terminal quadratic differential on @xmath209 is obtained as follows .
let @xmath210 be the natural parameter of @xmath17 on @xmath195 , i.e. @xmath211 ; let @xmath212 , where @xmath213 shrinks the vertical direction of @xmath210 by the multiplicative constant @xmath32 .
then the terminal quadratic differential @xmath207 is defined in the image of the natural parameter as @xmath214 . if @xmath215 then @xmath216 .
the metric @xmath217 is admissible for @xmath218 since @xmath219 for all @xmath32 and all @xmath220 . then @xmath221 which gives @xmath222 since @xmath223 is integrable
we estimate @xmath224 .
let @xmath225 be fixed .
denote by @xmath226 the path metric defined by integrating @xmath227 i.e. the @xmath17-metric .
let @xmath228 . for @xmath229
define @xmath230 .
given @xmath231 there exists @xmath232 such that @xmath233 denote by @xmath234 the set of vertical trajectories @xmath140 connecting @xmath191 with @xmath192 . the choice @xmath232 and
the fact that the vertical trajectories are geodesics for @xmath226 implies that @xmath235 . from now on we choose @xmath236 as above . for @xmath237 ,
define @xmath238 .
note that @xmath239\ ] ] which gives @xmath240 since @xmath33 is @xmath241-quasiconformal , we have @xmath242 define metric @xmath243 for @xmath244 and @xmath245 otherwise . then @xmath246 is admissible for the family @xmath247 .
thus @xmath248
we estimate @xmath249 .
note that @xmath250 is a compact metric space for the distance @xmath226 .
similar to the above @xmath251 by keith @xcite , we have that @xmath252 recall that a sequence @xmath253 of families of curves converges to a family @xmath87 if for each @xmath254 there exists a subsequence @xmath255 such that uniformly lipschitz parameterizations of @xmath256 converge to @xmath140 as functions when @xmath257 , and if the limit of each convergent subsequence is a curve in @xmath87 ( cf
. @xcite ) .
we establish that @xmath258 let @xmath259 be a sequence of uniformly lipschitz parametrizations of curves in @xmath260 with @xmath261 as @xmath262 that converges to @xmath263 .
then @xmath264 indeed , @xmath265 implies that @xmath266 for all @xmath267 large enough .
this contradicts @xmath268 . since @xmath269
, this implies @xmath270 or that @xmath140 is composed of several vertical trajectories that meet at a zero of @xmath17 .
the later curves are at most countable and their modulus is zero , so we can ignore them . since @xmath271 by our choice of @xmath229 ,
we obtain ( [ eq : lim_curves ] ) .
then ( [ eq : g_g ] ) , ( [ eq:>eta ] ) , ( [ eq:>r+1 ] ) and ( [ eq : lim_curves ] ) imply that @xmath272 we prove that @xmath234 can be replaced by @xmath182\times [ c , d])$ ] in ( [ eq : g_va ] ) .
note that ( [ eq : g_va ] ) is true for all @xmath191 and @xmath192 . choose a sequence @xmath273 and @xmath274 such that @xmath275\subset \mathbb{s}^1 $ ] and @xmath276\subset \mathbb{s}^1 $ ] as @xmath277 in the hausdorff topology on closed subsets of @xmath278 .
denote by @xmath279 the subset of @xmath4 corresponding to @xmath273 and @xmath274 .
define @xmath280,[c',d']):=\gamma_v([a',b'],[c',d'])\cap \mathbb{d}_k'.\ ] ] we claim that @xmath281,[c',d']))=0.\ ] ] indeed , let @xmath282 be the lower bound on the distance @xmath226 between @xmath273 and @xmath274 over all @xmath283 .
then @xmath284 is admissible for @xmath285 .
let @xmath286 be the union of the ( complete ) vertical trajectories in @xmath4 that connect @xmath273 and @xmath274 and do not connect @xmath287 $ ] and @xmath288 $ ] .
then @xmath289 for all @xmath283 ( since we can choose @xmath273 and @xmath274 such that @xmath290 ) .
we claim that @xmath291 .
assume that a horizontal trajectory @xmath140 belongs to the union that makes @xmath286 .
then there exists either a euclidean neighborhood of @xmath287 $ ] or a euclidean neighborhood of @xmath288 $ ] in @xmath278 such that @xmath140 is disjoint from this neighborhood .
there exists @xmath292 such that @xmath140 does not intersect either @xmath293 or @xmath294 .
thus @xmath140 does not belong to @xmath295 and @xmath291 .
this gives @xmath296 as @xmath277 and ( [ eq : conv_l_a ] ) follows . from ( [ eq : g_va ] ) and ( [ eq : conv_l_a ] ) we get @xmath297,[c',d'])).\ ] ] by keith @xcite , we have that @xmath298,[c',d']))\leq { \mathrm{mod } } ( \limsup_{k\to\infty } \gamma_v^k([a',b'],[c',d']))\ ] ] where @xmath299,[c',d'])$ ] is for the euclidean metric on @xmath278 . for every point of @xmath140 which is not a zero of @xmath17
, there exists an open subarc of @xmath140 containing the point that is a part of a vertical trajectory of @xmath17 because @xmath300 are vertical trajectories .
moreover the limit @xmath140 has one endpoint in @xmath287 $ ] and the other endpoint in @xmath288 $ ] because @xmath300 has one endpoint on @xmath273 and one endpoint on @xmath274 , and @xmath273 converges to @xmath287 $ ] and @xmath274 converges to @xmath288 $ ] .
therefore every limit @xmath140 is a vertical trajectory that necessarily belongs to @xmath301,[c',d'])$ ] or it is composed of several vertical trajectories meeting at zeros of @xmath17 .
the later family is countable and of zero modulus and without loss of generality we ignore it
. therefore @xmath302,[c',d'])).\ ] ] we choose sequences @xmath303\supset [ a , b]$ ] and @xmath304\supset [ c , d]$ ] on finite distance from @xmath305 such that @xmath306 , @xmath307 , @xmath308 and @xmath309 as @xmath277 .
the inequality ( [ eq : ineq_vert ] ) holds for these sequences and we need to prove that it holds for @xmath182,[c , d])$ ] as well .
it is enough to prove that @xmath310,[c_k',d_k'])-\gamma_v([a , b],[c , d]))\to 0.\ ] ] let @xmath311 be the union of vertical trajectories in @xmath312,[c_k',d_k'])-\gamma_v([a , b],[c , d])$ ] and note @xmath313 . define @xmath314 for @xmath315 and @xmath245 otherwise , where @xmath316 is the length of the vertical trajectory through @xmath317 with respect to the metric @xmath226 . then @xmath86 is allowable metric for the family @xmath312,[c_k',d_k'])-\gamma_v([a , b],[c , d])$ ] .
we have @xmath310,[c_k',d_k'])-\gamma_v([a , b],[c , d]))\leq\iint_{\mathbb{d}_k}\frac{1}{l_v(z)^2}|\varphi ( z)|dxdy.\ ] ] we claim that @xmath316 has a positive lower bound in @xmath311 . indeed , since intervals @xmath303 $ ] and @xmath304 $ ] are disjoint and decreasing
, their distance in @xmath226 metric is positive which implies that any vertical trajectories connecting them must have lengths bounded below by a positive constant .
thus @xmath318 is bounded above .
then @xmath319 implies that @xmath320 as @xmath277 .
the proof is finished .
[ thm : main ] let @xmath87 be the family of rectifiable arcs in @xmath4 with one endpoint in @xmath20\subset \mathbb{s}^1 $ ] and the other endpoint in @xmath22\subset \mathbb{s}^1 $ ]
. denote by @xmath33 the teichmller map of @xmath4 that shrinks the vertical trajectories of @xmath17 by the multiplicative constant @xmath32 . then @xmath321,[c , d]))\ ] ] where @xmath182,[c , d])$ ] is the set of vertical trajectories with one endpoint in @xmath20 $ ] and the other endpoint in @xmath22 $ ] .
we keep the notation as in the proof of proposition [ prop : upper_bound ] .
since @xmath182,[c , d])\subset \gamma$ ] , it follows that @xmath322,[c , d]))\leq { \mathrm{mod}}(\gamma ) $ ] . because @xmath182,[c , d])$ ] consists of only vertical trajectories
, it follows that @xmath323,[c , d])))={\mathrm{mod}}(\gamma_v([a , b],[c , d])).\ ] ] thus @xmath324,[c , d]))\leq\liminf_{\epsilon\to 0^{+ } } \epsilon\cdot { \mathrm{mod}}(t_{\epsilon}(\gamma ) ) .\ ] ] the opposite inequality is obtained in proposition [ prop : upper_bound ] and theorem follows .
.2 cm we give an equivalent definition of @xmath322,[c , d]))$ ] .
.2 cm [ prop : vertical_mod_measure ] let @xmath17 be an integrable holomorphic quadratic differential on the unit disk @xmath4
. then @xmath324,[c , d]))=\int_i \frac{1}{l(z)}|re(\sqrt{\varphi ( z)}dz)|\ ] ] where @xmath64 is at most countable set of horizontal arcs that intersects each trajectory of @xmath182,[c , d])$ ] in one point and no other vertical trajectories up to countably many of them , and @xmath325 is the length of the vertical trajectory through @xmath317 .
the metric @xmath326 is allowable for the family @xmath182,[c , d])$ ] and thus @xmath322,[c , d]))\leq \int_i \frac{1}{l(z)}|\sqrt{\varphi ( z)}dz|$ ] .
we claim that @xmath246 is extremal metric for the family @xmath182,[c , d])$ ] which proves that we have equality above . using beurling s criterion of sufficiency for extremal metrics @xcite , we need to show that if @xmath327 for all @xmath328,[c , d])$ ] and some @xmath329
then we have @xmath330 . by transferring the integration to the natural parameter
, we get that @xmath331 are subsets of vertical lines which implies @xmath332 and @xmath333 .
note that @xmath334 is independent of @xmath335 . then @xmath336 and multiplying with @xmath337 and an integration in the @xmath338 direction gives the desired inequality ( cf .
@xcite ) . define a measured lamination @xmath23 as follows .
the support of @xmath23 is a geodesic lamination @xmath19 obtained by taking geodesics in @xmath4 which are homotopic to the vertical trajectories of @xmath17 relative their endpoints on @xmath9 , i.e. a geodesic in the support @xmath19 of @xmath23 has endpoints equal to a vertical trajectory of @xmath17 . for a box of geodesics @xmath20\times [ c , d]$ ] , define @xmath339\times [ c , d])={\mathrm{mod}}(\gamma_v([a , b],[c , d])).\ ] ] note that @xmath23 is a measure on the space of geodesics ( i.e. it is countable additive ) by the above integration formula for @xmath322,[c , d])).$ ] also note that @xmath340,[c , d]))$ ] is not countably additive ( since moduli are only countably subadditive ) and hence it does not define a measure on @xmath18 .
.2 cm [ prop : common_endpoints ] let @xmath23 be the measured geodesic lamination corresponding to an integrable holomorphic quadratic differential @xmath17 on @xmath4 as above
. then @xmath30)=0\ ] ] for all @xmath31 and @xmath22\subset \mathbb{s}^1 $ ] with @xmath341 $ ] .
we recall that @xmath4 is covered by countably many mutually disjoint open strips @xmath159 up to countably many vertical trajectories .
assume on the contrary that @xmath342)>0 $ ] .
then there exists an open strip @xmath343 such that @xmath344 where @xmath345 is the open arc on a horizontal trajectory and @xmath346)$ ] . by the definition
, @xmath347 is the horizontal measure in the natural parameter of @xmath17 of the vertical trajectories of @xmath17 intersecting @xmath345 . for @xmath348 ,
let @xmath325 be the length of the vertical trajectory through @xmath317 .
since @xmath17 is integrable , we have that @xmath349 which implies that @xmath350 for a.e . @xmath348 .
let @xmath351 be such that there exists @xmath352 with @xmath353 for a linear order on @xmath345 , and @xmath354 and @xmath355 finite .
let @xmath356 be the maximal vertical rays starting at @xmath357 respectively that have @xmath358 as their common endpoint .
note that vertical rays @xmath359 and @xmath360 do not intersect @xmath361 except at their initial points because any two points in @xmath4 can be joined by at most one geodesic arc in the metric @xmath227 ( cf .
* theorem 14.2.1 , page 72 ) ) .
let @xmath362 $ ] be the subarc of the vertical trajectory between @xmath363 and @xmath364 .
then @xmath365\cup\ { a\}$ ] is the boundary of a simply connected domain @xmath366 inside @xmath4 .
we claim that @xmath366 is a jordan domain .
indeed , since @xmath359 , @xmath360 and @xmath362 $ ] are simple geodesic arcs which meet only at their endpoints , it follows that @xmath367 $ ] is a jordan arc .
we parametrize it by a homeomorphism @xmath368 $ ] and extend @xmath369 .
then @xmath370 is a bijection of @xmath9 and @xmath371\cup\ { a\}$ ] .
moreover , @xmath370 is continuous at @xmath6 since @xmath359 and @xmath360 accumulate to @xmath358 and therefore @xmath372 is a jordan curve . for @xmath373 $ ] , let @xmath374 be the ray of the vertical trajectory with the initial point @xmath317 that starts in the direction of @xmath366 .
then @xmath374 never leaves @xmath366 because it can not intersect its boundary except at @xmath317 .
moreover , the ray @xmath374 can not contain critical points of @xmath17 .
indeed , if it does contain a critical point then there exist two vertical rays starting at the critical point which make a geodesic and whose both accumulation points on @xmath9 are equal to @xmath358 .
however , a geodesic must have two different accumulation points ( cf .
* theorem 19.4 and theorem 19.6 ) ) which gives a contradiction .
therefore every vertical trajectory in @xmath366 is non - critical and its full extension accumulates at @xmath375 and intersects @xmath362 $ ] in exactly one point .
therefore , @xmath366 is foliated by @xmath374 for @xmath376 .
consider the conformal mapping from @xmath366 into @xmath16 using the natural parameter @xmath377 .
since @xmath366 is simply connected and without zeroes , the natural parameter is conformal on @xmath366 .
caratheodory s theorem ( cf .
@xcite ) gives that @xmath148 homeomorphically maps the boundary @xmath372 of @xmath366 onto the prime ends of @xmath378 .
since @xmath379 and @xmath380 have finite lengths , it follows that the endpoints @xmath381 and @xmath382 of vertical lines @xmath383 and @xmath384 are different in @xmath385 . the arcs @xmath379 and @xmath380 define degenerate prime ends , namely prime ends whose imprints are @xmath381 and @xmath382
. therefore the prime ends are different since @xmath386 ( cf . figure 3 ) .
this is impossible since @xmath148 maps @xmath358 onto both prime ends . contradiction .
thus we obtained that @xmath342)=0 $ ] . putting the above statements together and using the fact that the asymptotics of the liouville currents can be replaced by the asymptotics of the moduli of curves ( cf .
lemma [ lem : mod_liouville_measure ] ) gives [ thm : weak*convergence ] let @xmath17 be an integrable holomorphic quadratic differential on @xmath4 and let @xmath33 be the teichmller mapping that shrinks the vertical trajectories of @xmath17 by a multiplicative constant @xmath32 .
the teichmller ray @xmath387 for @xmath32 has a unique limit point @xmath51 $ ] on thurston s boundary @xmath10 of @xmath3 as @xmath36 , where @xmath51 $ ] is the projective class of a bounded measured lamination @xmath23 corresponding to @xmath17 . the convergence @xmath388 $ ] as @xmath39 in the weak * topology on measures follows immediately from theorem [ thm : main ] , proposition [ prop : common_endpoints ] and lemma [ lem : mod_liouville_measure ] .
it remains to be proved that @xmath23 is thurston bounded .
note that by the definition the measured lamination @xmath23 is independent under multiplication of @xmath17 by positive constants .
let @xmath20\times [ c , d]$ ] be such that its liouville measure satisfies @xmath78\times [ c , d])=\log 2.\ ] ] denote by @xmath389,[c , d])$ ] the family of all rectifiable arcs in @xmath4 that have one endpoint in @xmath20 $ ] and other endpoint in @xmath22 $ ] .
then @xmath390,[c , d]))\leq const\ ] ] for all @xmath82\times [ c , d])=\log 2.$ ] since @xmath182,[c , d])\subset \gamma ( [ a , b],[c , d])$ ] , we have that @xmath339\times [ c , d])={\mathrm{mod}}(\gamma_v ( [ a , b],[c , d]))\leq const\ ] ] and @xmath391 .
let @xmath392 and @xmath80 be geodesic currents on the space of geodesics @xmath393 of the hyperbolic plane @xmath2 .
namely , @xmath392 and @xmath80 are positive radon measures on @xmath18 .
we say that @xmath394 converges to @xmath80 in the _ uniform _ weak * topology ( cf .
@xcite ) if for every continuous @xmath395 with support in the standard box @xmath396\times [ -1,-i]$ ] we have @xmath81\times [ c , d]}|\int_{[1,i]\times [ -1,-i]}f[d\gamma_{[a , b]\times [ c , d]}^{*}(\alpha_n-\alpha ) ] |\to 0\ ] ] as @xmath262 , where the supremum is over all boxes @xmath20\times [ c , d]$ ] of liouville measure @xmath397 and @xmath398\times [ c , d]}$ ] is the mbius map taking the standard box onto @xmath20\times [ c , d]$ ] .
we note that as @xmath20\times [ c , d]$ ] runs through all boxes of liouville measure @xmath397 , @xmath398\times [ c , d]}$ ] runs through all mbius maps of @xmath4 , which implies that the above supremum can be taken over the space @xmath399 of all mbius maps that preserve the unit disk @xmath4 .
moreover , since any continuous @xmath400\times [ c_0,d_0]\to\mathbb{r}$ ] with @xmath401\times [ c_0,d_0])=\log 2 $ ] can be pulled back to a continuous @xmath402\times [ c_0,d_0]}:[1,i]\times [ -1,-i]\to\mathbb{r}$ ] and since the above supremum is over all mbius maps , we do not need to restrict to continuous functions with supports on the standard box , but rather to continuous maps with supports in any box with liuoville measure @xmath397 .
in addition , if a continuous @xmath403 has a compact support then it can be written as a finite sum of continuous functions with supports in boxes of liouville measures @xmath397 . therefore a sequence of geodesic currents @xmath404 converges in the _ uniform _ weak * topology to @xmath80 if for every continuous function @xmath403 with compact support @xmath405|\to 0\ ] ] as @xmath262 .
this definition is equivalent to the first definition using boxes of liouville measure @xmath397 .
assume that @xmath394 converges to @xmath80 in the weak * topology .
below we formulate a sufficient condition guaranteeing that @xmath394 does not converge to @xmath80 in the uniform weak * topology .
given @xmath406 , assume that there exist @xmath407 and a sequence of boxes @xmath408\times [ c_k , d_k]$ ] and sub - boxes @xmath409\times [ c_k',d_k']$ ] compactly contained in the interior of @xmath410 such that @xmath411 @xmath412 @xmath413\times [ c_k , d_k]),\mathcal{l}([b_k',b_k]\times [ c_k , d_k]),\ \ \ \ \ \ \ \ \ \\ \mathcal{l } ( [ a_k , b_k]\times [ c_k , c_k ' ] ) , \mathcal{l}([a_k , b_k]\times [ d_k',d_k])\}\geq\delta > 0 , \end{split}\ ] ] @xmath414 for some @xmath415 with @xmath257 as @xmath277 , and @xmath416 as @xmath277 , where @xmath417 , @xmath418 , @xmath419 and @xmath420 are independent of @xmath283 and @xmath421 .
we now establish that @xmath394 does not converge to @xmath80 in the uniform weak * topology if the above conditions are satisfied .
let @xmath422\times [ c , d]$ ] be a fixed box with @xmath423 and let @xmath424\times [ c',d']$ ] be a box compactly contained in the interior of @xmath425 such that @xmath426\times [ c , d])=\mathcal{l}([b',b]\times [ c , d])=\ \ \ \ \\ \mathcal{l } ( [ a , b]\times [ c',c])=\mathcal{l}([a , b]\times [ d',d])=\delta . \end{split}\ ] ] let @xmath427 be such that @xmath428 . then
( [ eq : dist - bdry ] ) and ( [ eq : central ] ) imply that @xmath429 .
let @xmath395 be a continuous functions such that the support of @xmath370 is contained in @xmath425 , @xmath430 and @xmath431 .
then by ( [ eq : lower - bound ] ) and ( [ eq : zero - measure ] ) we have @xmath432\geq c_3-\alpha ( q_{k})>c_3/2>0\ ] ] when @xmath415 is large , which implies that @xmath433 does not converge in the uniform weak * topology to @xmath80 .
since the uniform weak * convergence implies the weak * convergence and since @xmath394 converges in the weak * topology to @xmath80 , it follows that @xmath394 does not converge to any geodesic current in the uniform weak * topology .
we find an example of an integrable holomorphic quadratic differential @xmath17 on the unit disk @xmath4 such that the corresponding teichmller ray @xmath33 does not converge in the uniform weak * topology to @xmath23 while theorem [ thm : weak*convergence ] established that it does converge to @xmath23 in the weak * topology .
the differential @xmath17 is constructed by taking the pull back of @xmath434 on the domain @xmath435 in the lemma below under the riemann mapping . for simplicity of notation , we denote by @xmath436 the boundary map of the teichmller geodesic ray @xmath33 .
the above criterion is used for the family of liouville currents @xmath437 when @xmath39 and the weak * limit @xmath23 .
the conditions ( [ eq : constant - measure ] ) , ( [ eq : bounded - measure ] ) , ( [ eq : central ] ) , ( [ eq : lower - bound ] ) and ( [ eq : zero - measure ] ) are replaced by equivalent conditions in terms of the moduli of the families of curves connecting two intervals on @xmath9 defining the box of geodesics .
[ lemma : counterexample ] there is a domain @xmath438 of finite area with the following properties .
there exist constants @xmath439 , a sequence of arcs @xmath440 , [ c_k , d_k]$ ] and sub - arcs @xmath303,[c_k',d_k']$ ] on @xmath9 and a sequence @xmath441 approaching @xmath135 such that with notations as above we have * @xmath442 , [ c_k , d_k];\mathbb{d})\leq c_1 ' , \forall k\in\mathbb{n}$ ] * @xmath443 , [ c_k',d_k'];\mathbb{d})\geq c_2 ' , \forall k\in\mathbb{n}$ ] * @xmath444 , [ c_k , d_k];\mathbb{d}),{\mathrm{mod}}([b_k',b_k ] , [ c_k , d_k];\mathbb{d } ) , \\ { \mathrm{mod}}([a_k , b_k ] , [ c_k , c_k'];\mathbb{d } ) , { \mathrm{mod}}([a_k , b_k ] , [ d_k',d_k];\mathbb{d})\}\geq\delta ' > 0,$ ] * @xmath445\times [ c_k , d_k])\to0 , \mbox { as } k\to\infty$ ] * @xmath446 ) , t_{\epsilon_k}^{\varphi}([c_k',d_k']);t_{\epsilon_k}^{\varphi}(\mathbb{d}))\geq c_3'.$ ] we would like to emphasize again that in the lemma above @xmath17 denotes the quadratic differential which is the pullback of @xmath434 under the riemann map of @xmath435 and @xmath447 is the corresponding teichmller mapping .
the boxes @xmath440\times [ c_k , d_k]$ ] and @xmath303\times [ c_k',d_k']$ ] on @xmath9 under the riemann mapping correspond to boxes in @xmath435 , and this correspondence is implicitly assumed . below we will define the domain @xmath435 as well as a sequence of continua @xmath448,@xmath449 , which are the preimages of the intervals @xmath440,[a_k',b_k'],[c_k , d_k],[c_k',d_k']\subset \mathbb{s}^1 $ ] under the riemann mapping of @xmath435 . in particular @xmath450 and @xmath451 .
moreover , instead of estimating the moduli of the curve families in the unit disc @xmath4 , we will obtain the estimates in @xmath435 . to simplify the notation we let @xmath452 furthermore , denoting the two nonempty components of @xmath453 and @xmath454 by @xmath455 and @xmath456 , @xmath457 , respectively , we let @xmath458 just as before , given two continua @xmath459 we denote by @xmath460 the family of vertical curves connecting @xmath97 and @xmath98 in @xmath435 . by conformal invariance of the modulus and theorem [ thm : main ] conditions @xmath461 are equivalent to the following : * @xmath462 * @xmath463 * @xmath464 * @xmath465 * @xmath466 next , we construct the domain @xmath435 and prove properties @xmath467 . for @xmath468 and @xmath469 let @xmath470 define @xmath471 ^ 2
\setminus \bigcup_{k=1}^{\infty}\bigcup_{j=0}^{2^k } l_{k , j}.\ ] ] .
the bold interval is @xmath472 , while the part of @xmath473 above it is @xmath474.,width=377 ] now , for @xmath475 let @xmath476 \quad & \mbox { and } \quad e_k=\bigcup_{j=0}^{2^k } l_{k , j } \cup \left\{(x,1 ) : x\in f_k \right\},\\ f_k'=\frac{1}{2}f_k \quad & \mbox { and } \quad e_k'=\bigcup_{j=\frac{1}{4}2^k}^{\frac{3}{4}2^k } l_{k , j } \cup \left\{(x,1 ) : x\in f_k ' \right\},\end{aligned}\ ] ] where @xmath477 denotes the interval with the same center as @xmath478 but half the length .
* proof of ( _ _ a__@xmath479 ) . * since @xmath480 for every @xmath475 , by lemma [ lemma : mod - reldist ] we have @xmath481 * proof of ( _ _ b__@xmath479 ) . * to estimate @xmath482 from below we will use conjugate families .
recall that if continua @xmath483 then the family of curves separating @xmath97 and @xmath98 in @xmath435 is called the family conjugate to @xmath484 .
we will denote by @xmath485 the family conjugate to @xmath486 .
the modulus of @xmath485 may be found as follows , see @xcite @xmath487 thus , to estimate @xmath482 from below we can instead estimate @xmath488 from above .
note , that every curve @xmath489 contains a subcurve @xmath125 connecting the two components of @xmath490 in the rectangle @xmath491 $ ] , see fig .
[ fig : zoom ] .
therefore , @xmath492 where by @xmath493 we denote the family of curves connecting the two components of @xmath494 in the rectangle @xmath491 $ ] .
next we estimate @xmath495 using the following result .
[ fig : zoom ] from below .
every curve @xmath489 separating @xmath496 from @xmath497 in @xmath435 contains a subcurve @xmath498 which connects the two components of @xmath499 within the grey rectangle @xmath500.,title="fig:",width=377 ] [ lemma : slits ] let @xmath501 , @xmath502 and @xmath503 .
denote by @xmath504 the domain @xmath505,\end{aligned}\ ] ] where @xmath506 and by @xmath507 the family of curves in @xmath504 connecting the vertical intervals @xmath508 to @xmath509 ( see figure [ fig : vertslits ] ) .
then @xmath510 the first estimate follows from the fact that @xmath507 contains the family connecting the vertical sides in the rectangle @xmath511\times [ 0,c]$ ] . to obtain the upper bound consider the rectangle @xmath512 and define @xmath513 we next show that @xmath514 is admissible for @xmath507 . for that , let @xmath515 and let ( cf . figure 6 ) @xmath516 next , we show that @xmath517 indeed , if @xmath518 then , since @xmath519 is a connected curve connecting the vertical sides of the rectangle @xmath520 , we have @xmath521 . on the other hand if @xmath522 then there are two connected components @xmath523 of @xmath524 which connect the vertical intervals @xmath525 and @xmath526 to the horizontal interval @xmath527 in @xmath504 , respectively . since the distance between the aforementioned vertical intervals and the horizontal interval is at least @xmath528 we obtain @xmath529 [ fig : vertslits ] thus we have @xmath530 and @xmath514 is admissible for @xmath507 . therefore we can estimate the modulus of @xmath507 as follows .
@xmath531 using the lemma we see that @xmath532 for @xmath475 . which proves @xmath533 .
* proof of ( _ _ c__@xmath479 ) .
* we start by estimating the modulus of @xmath534 .
note that @xmath535 while @xmath536 and @xmath537 are the parts of @xmath538 above @xmath539 and @xmath540 , respectively . just like in the proof of @xmath533 we use lemma [ lemma :
slits ] to obtain the following estimate @xmath541 ) \\ & = \frac{1}{{\mathrm{mod}}(f_k^1,e_k^1,f_k^1\times[0,1])^t } \geq \frac{1}{\frac{2^{-k}}{2^{-k}/4 } + \frac{1}{2 ( 2^k/4 ) } } \geq \frac{1}{5 } , \end{split}\end{aligned}\ ] ] for @xmath475 .
the same way we also obtain @xmath542 .
next , we estimate @xmath543 as follows @xmath544)= \frac{1}{{\mathrm{mod}}(f_k^1,e_k^2,f_k^1\times[0,1])^t}. \end{split}\end{aligned}\ ] ] but @xmath545)^t = ( f_k'',e_k'',f_k^1\times[0,1])\end{aligned}\ ] ] where @xmath546 and @xmath547 are the two components of @xmath548 , or
@xmath549\cup e_k^1\cup e_k',\\ f_k '' & = [ 2^{-k+1},(1+i)2^{-k+1}]\cup f_k'\cup f_k^2.\end{aligned}\ ] ] since @xmath550 and @xmath551 we have @xmath552 therefore by lemma [ lemma : mod - reldist ] we have @xmath553)\leq \pi(1 + 4)^2 = 25\pi,\end{aligned}\ ] ] and we finally obtain @xmath554 in the same way we can show that @xmath555 and thus prove @xmath556 . * proof of ( _ _ d__@xmath479 ) .
* as was shown before we have , @xmath557 where in this case @xmath25 is the euclidean length of the vertical trajectory passing through @xmath558 and integration is with respect to the lebesgue measure .
thus , since @xmath559 for almost every @xmath560 we obtain @xmath561 * proof of ( _ _ e__@xmath479 ) . *
just like above , let @xmath493 be the family of curves connecting the two vertical intervals in @xmath562 ( namely , the two components of the boundary of @xmath563 ) ) .
then @xmath564 overflows @xmath493 , and the same way we also have @xmath565 overflows @xmath566 and therefore for every @xmath567 we have @xmath568 next , we let @xmath569 and estimate @xmath570 from above . considering the conformal mapping @xmath571 and using lemma [ lemma : slits ] we obtain that @xmath572 now , since @xmath573 , we obtain @xmath574
let @xmath17 be an integrable holomorphic quadratic differential on @xmath4 ( i.e. a holomorphic function @xmath143 such that @xmath575 ) .
let @xmath576 be the space of all integrable holomorphic quadratic differentials on @xmath4 .
given @xmath577 , we defined a corresponding bounded measured lamination @xmath339\times [ c , d])={\mathrm{mod}}(\gamma_v ( [ a , b],[c , d]))\ ] ] or equivalently @xmath339\times [ c , d])=\int_i \frac{1}{l(z)}|\sqrt{\varphi ( z)}dz|\ ] ] where @xmath64 is transverse arc to @xmath578,[c , d])$ ] .
it follows that if @xmath45 then @xmath47 .
therefore we obtain a map from the space @xmath579 of projective integrable holomorphic quadratic differentials to the space of projective bounded measured laminations @xmath10 , @xmath580 we prove that @xmath581 is injective .
we assume that @xmath584 and need to prove that @xmath585 for some @xmath45 . since @xmath586 we have that their geodesic laminations supports @xmath587 and @xmath588 are the same .
in other words each leaf of the vertical foliation @xmath17 is homotopic to a leaf of the vertical foliation of @xmath589 relative their two endpoints on the unit circle , and vice versa .
additionally , assume that the corresponding leaves of the vertical foliations are not only homotopic but that they are equal to each other .
in other words , the vertical foliations of @xmath17 and @xmath589 are equal .
if @xmath590 is a regular point of both @xmath17 and @xmath589 , denote by @xmath210 and @xmath591 the corresponding natural parameters in a regular neighborhood @xmath366 of @xmath305 .
then @xmath592 is a conformal mapping from @xmath593 onto @xmath594 that maps vertical lines onto vertical lines .
it follows then that @xmath595 for some @xmath596 .
thus @xmath597 and we set @xmath598 .
we obtained that for each regular point @xmath305 of @xmath17 and @xmath589 there exist a neighborhood @xmath599 and a constant @xmath600 such that @xmath585 in @xmath366 .
since the set of regular points of @xmath17 and @xmath589 is connected and dense in @xmath4 then @xmath601 in @xmath4 and the proof is finished in this case .
it remains to prove that the vertical foliations of @xmath17 and @xmath589 are the same under the assumption that @xmath586 .
let @xmath602 be a family of mutually disjoint vertical strips with open transverse horizontal arcs @xmath160 that covers @xmath4 up to countably many vertical trajectories ( cf .
the metric on the horizontal arcs @xmath160 is induced by @xmath603 and we isometrically identify @xmath160 with @xmath604 , where @xmath605 is the length of @xmath160 .
the variable in @xmath604 is @xmath338 and the integration with respect @xmath606 corresponds to integration with respect @xmath603 in @xmath4 .
the arc @xmath604 is a horizontal arc in the natural parameter @xmath607 for @xmath608 . for @xmath160 ,
let @xmath609 be the substrip of @xmath159 of vertical trajectories going through @xmath610 .
the area of @xmath609 is @xmath611 where @xmath612 is the vertical trajectory of @xmath17 through the point @xmath613 and @xmath614 is the length in the @xmath603 metric .
the modulus of the vertical trajectories in @xmath609 is @xmath615 if necessary , we multiply @xmath589 by a positive constant such that @xmath616 . since the supports of @xmath23 and @xmath617 are the same , to each @xmath609 there corresponds a vertical strip @xmath618 of vertical trajectories @xmath619 of @xmath589 with the same endpoints on @xmath9 as @xmath612 .
note that @xmath619 does not necessarily pass through @xmath620 or even intersects @xmath160 .
[ lem : weak ] let @xmath160 be a transverse horizontal arc to a vertical strip @xmath159 isometrically identified with @xmath604 in the natural parameter of @xmath17 . then for a.e .
@xmath623 , we have @xmath624 ^ 2},\ ] ] where @xmath614 is the @xmath17-length and @xmath625 is the horizontal trajectory of @xmath589 whose endpoints agree with the endpoints of @xmath626 . for @xmath623 and small @xmath567 we denote @xmath627 } \}.\ ] ] note that @xmath628 for @xmath567 small enough by the continuity of @xmath589 .
it is possible that @xmath629 if all vertical trajectories of @xmath589 close to @xmath625 have infinite @xmath17-length .
the metric @xmath630 is admissible for @xmath631)$ ] , where @xmath632 if @xmath629 .
since , @xmath633 is non - increasing it has a limit as @xmath634 . in fact , we have @xmath635{}l^{\varphi}(v_{\beta_i}^{\varphi'}(t)).\ ] ] to see this , note first that @xmath636 and we only need to estimate the limit from below .
assume first that @xmath637 .
fix @xmath638 and choose points @xmath639 , so that @xmath640 where @xmath641 is the distance between @xmath642 and @xmath643 in the @xmath17 metric , i.e. the metric induced by @xmath603 .
we want to show that for small @xmath644 the curves @xmath645 have lengths at least @xmath646 .
since the set of vertical trajectories @xmath159 foliates a neighborhood of @xmath626 then @xmath647 must foliate a neighborhood of @xmath625 because the separation property of the vertical trajectories of @xmath17 and @xmath589 is a topological property of their endpoints . by choosing small @xmath200
, we get that a subarc of @xmath648 is within small euclidean distance to the subarc of @xmath625 between @xmath649 and @xmath650 for all @xmath651 .
since @xmath17 is continuous , it follows that for @xmath200 small enough , each @xmath648 for @xmath652 has points @xmath653 on the @xmath17-distance less than @xmath654 from @xmath655 , respectively .
therefore by ( [ eq:1 ] ) we have @xmath656 thus @xmath657 for all @xmath652 which implies that @xmath658 because @xmath406 is arbitrary .
assume now that @xmath659 .
if @xmath629 for some @xmath567 then @xmath660 . we consider the case when @xmath661 for all @xmath567 and need to prove that for every @xmath237 there exist @xmath200 such that @xmath662 for all @xmath663 . choose points @xmath639 , so that @xmath664 where @xmath641 is the distance between @xmath642 and @xmath643 in the @xmath17-metric .
for small @xmath200 and all @xmath663 , a subarc of @xmath665 is within small euclidean distance to the subarc of @xmath625 between @xmath649 and @xmath650 .
since @xmath17 is continuous , it follows that for @xmath200 small enough , each @xmath648 for @xmath652 has points @xmath653 on the @xmath17-distance less than @xmath666 from @xmath655 , respectively .
therefore by ( [ eq:2 ] ) we have @xmath667 and then @xmath668 .
@xmath669)}{{\varepsilon}}\\ & \leq & \limsup_{{\varepsilon}\to 0^{+ } } \frac{a_{\beta_i}^{\varphi'}(x+{\varepsilon})- a_{\beta_i}^{\varphi'}(x)}{{\varepsilon}l^2_x({\varepsilon } ) } = \frac{\frac{d}{dx}a_{\beta_i}^{\varphi'}(x ) } { [ l^{\varphi}(v_{\beta_i}^{\varphi'}(x))]^2},\end{aligned}\ ] ] note that @xmath671 and @xmath672 , and equation ( [ eq : vv ] ) imply that @xmath673 for every @xmath623 .
by the previous lemma , by lebesque s differentiation theorem and by absolute continuity of @xmath674 we have for a.e .
@xmath623 @xmath675 ^ 2}=\frac{1}{[l^{\varphi}(v_{\beta_i}^{\varphi}(x))]^2}\lim_{{\varepsilon}\to 0}\frac{1}{{\varepsilon}}\int_x^{x+{\varepsilon}}l^{\varphi}(v_{\beta_i}^{\varphi}(t))dt = \\
\frac{1}{l^{\varphi}(v_{\beta_i}^{\varphi}(x ) ) } = \frac{d}{dx}m_{\beta_i}^{\varphi}(x)= c_1\frac{d}{dx}m_{\beta_i}^{\varphi'}(x)\leq \frac { c_1\frac{d}{dx}a_{\beta_i}^{\varphi'}(x)}{[l^{\varphi}(v_{\beta_i}^{\varphi'}(x))]^2}. \end{split}\end{aligned}\ ] ] note that that @xmath678 is absolutely continuous in @xmath338 , since @xmath679 .
thus @xmath680 for a.e .
@xmath623 . since @xmath681 and @xmath682
, we necessarily have equality for each term @xmath160 and for a.e . @xmath623
. therefore , by ( [ ineq : mod - derivative ] ) we have @xmath686 ^ 2}\leq \frac{c_1\frac{d}{dx}a_{\beta_i}^{\varphi'}(x)}{[l^{\varphi}(v_{\beta_i}^{\varphi'}(x))]^2},\ ] ] where the numerators are equal for a.e .
@xmath687 . in particular , @xmath688 and
thus @xmath689 for a.e . @xmath338 . by the uniqueness of geodesics in the @xmath17
metric connecting two boundary points of simply connected domains ( cf .
@xcite ) and since vertical trajectories foliate @xmath4 , we obtain that all vertical trajectories of @xmath17 and @xmath589 are the same .
given an integrable holomorphic quadratic differential @xmath17 on the unit disk , we denote by @xmath54 the measured lamination whose support @xmath19 is homotopic to the leaves of the vertical foliation of @xmath17 and the transverse measure is given by @xmath690 , where @xmath64 is a horizontal arc intersecting the leaves of the vertical foliation corresponding to the leaves of @xmath54 .
we prove that @xmath54 is thurston bounded .
let @xmath691 be the the set of all vertical trajectories of @xmath17 whose @xmath17-length is @xmath692 .
let @xmath693 be the the set of all vertical trajectories of @xmath17 whose @xmath17-length is @xmath694 .
let @xmath695 and @xmath696 .
let @xmath20\times [ c , d]\subset ( \mathbb{s}^1\times \mathbb{s}^1)-diag$ ] be a box of geodesics with @xmath82\times [ c , d])=\log 2 $ ] .
this implies that @xmath697,[c , d];\mathbb{d})\leq c$ ] for some @xmath698 .
let @xmath64 be at most countable union of horizontal arcs of @xmath17 that intersects each vertical trajectory of @xmath17 in exactly one point .
then we have @xmath699 where @xmath338 is the real part of the natural parameter along @xmath64 . for a box of geodesics @xmath20\times [ c , d]$ ] , let @xmath700\times [ c , d]}$ ] be the set of vertical trajectories of length less than @xmath6 that connects @xmath20 $ ] and @xmath22 $ ] .
let @xmath701\times [ c , d]}$ ] be the subset of @xmath64 that intersects only vertical trajectories of @xmath700\times [ c , d]}$ ] .
then we have @xmath702,[c , d];\mathbb{d})\geq { \mathrm{mod}}(\mathcal{v}^ { < 1}_{[a , b]\times [ c , d]})=\\ \int_{i^ { < 1}_{[a , b]\times [ c , d]}}\frac{1}{l^{\varphi}(v^{\varphi}(z))}dx\geq \int_{i^ { < 1}_{[a ,
b]\times [ c , d]}}dx . \end{split}\ ] ] let @xmath21\times [ c , d]}$ ] be the subset of @xmath64 that intersects only the vertical trajectories of @xmath17 that connect @xmath20 $ ] to @xmath22 $ ] . since @xmath703\times [ c , d])=\int_{i_{[a , b]\times [ c , d]}}dx\leq \|\varphi\|_{l^1}+c$ ] , we have that @xmath704 .
a. fathi , f. laudenbach and v. po ' enaru , _
thurston s work on surfaces _ , translated from the 1979 french original by djun m. kim and dan margalit .
mathematical notes , 48 .
princeton university press , princeton , nj , 2012 .
o. lehto and k. i. virtanen , _ quasiconformal mappings in the plane .
_ , second edition
. translated from the german by k. w. lucas .
die grundlehren der mathematischen wissenschaften , band 126 .
springer - verlag , new york - heidelberg , 1973 .
j. p. otal , _ about the embedding of teichm " uller space in the space of geodesic h " older distributions .
handbook of teichm " uller theory _ ,
i , 223248 , irma lect .
phys . , 11 , eur .
, zrich , 2007 . | thurston s boundary to the universal teichmller space @xmath0 is the set of asymptotic rays to the embedding of @xmath0 in the space of geodesic currents ; the boundary is identified with the projective bounded measured laminations @xmath1 of @xmath2 .
we prove that each teichmller geodesic ray in @xmath0 has a unique limit point in thurston s boundary to @xmath0 unlike in the case of closed surfaces . |
the study of absorption line systems in the spectra of quasars ( qsos ) has been useful in detecting distant normal field galaxies situated close to the lines of sight of qsos .
conventionally , all such absorbers with velocity @xmath5 @xmath6 relative to the background qso are believed to be associated to the qso itself ( ` associated systems ' ) while those at larger velocity offset are believed to be entirely independent of background qso .
this general belief was questioned recently by the puzzling results on the abundance of strong absorber having equivalent width ( @xmath7 ) more than 1.0 : ( i ) by @xcite where they found @xmath8 time excess of strong absorber towards the @xmath9 ray burst ( grb ) sources relative to qso sight lines ( see also ) , and ( ii ) by , where they found similar excess by a factor of about 2 ( 3 @xmath10 confidence ) towards 45 blazar sight lines .
these counter - intuitive results , have inspired many alternative explanations , such as dust extinction towards qso sight lines which can lower the apparent incidence rate of absorbers , or gravitational lensing which can increase it toward grbs / blazars , but none have been found to explain the above discrepancies .
however the blazars , as a class , are believed to have relativistic jet pointed close to our line of sight . speculated that such powerful jets in the blazars are capable of sweeping sufficiently large column densities of gas ( up to @xmath11 ) and accelerating such clouds to velocities of order @xmath12 , thereby possibly accounting for the excess of absorption systems towards blazars in comparison with qsos .
however , such an excess in number of absorbers per unit redshift ( @xmath13 ) was not confirmed in the analysis of flat - spectrum radio quasars ( fsrqs ) by @xcite , though fsrqs also possess powerful jets similar to blazars , which they reconciled with the above hypothesis of relativistically ejected absorbing clouds , by suggesting that perhaps due to larger angle from the line of sight ( unlike blazars with smaller angle ) , these accelerated clouds might not intersect the line - of - sight in the case of fsrqs . using a larger sample size of 95 grb ( including 12 grb from @xcite ) , @xcite did not confirm the original enhancement found in the case of grb by @xcite , though a signature of mild excess of about 1.5 times was noticed for strong absorption systems , albeit with only a low confidence level of 90% .
the firm conclusion for jet based above excess still await the realistic numerical modelling of jet - ambient gas interaction especially for the excess seen towards blazars ( about a factor of 2 ) and cdqs ( about 10% ) @xcite vis - a - vis normal qsos .
however an alternative scenario , which could be more plausible , is the dust or radiation driven outflows ( e.g. * ? ? ?
for instance , if there is some contribution to @xmath13 of strong absorber from these outflows , then one will expect that agn luminosity should have statistical correlation with the velocity offset of the strong absorber relative to the background agn , which is usually defined by , @xmath14 where @xmath0 , @xmath15 is the emission redshift of the qso and @xmath16 is the absorption redshift of the system . in this letter
we report a correlation between the @xmath2 of strong absorbers and the bolometric luminosity ( @xmath1 ) of qsos , using the strong absorber catalogue by .
we also propose an explanation for this correlation which draws upon radiation driven outflow models . in
2 we describe the sample of strong absorbers and our selection criteria . in
3 we present our results and a theoretical model of radiation driven outflows . in
4 we study the fractional number counts of absorbers , and discuss our results in 5 .
we consider a sample of 10367 strong absorbers with equivalent width @xmath17 belonging to 9144 qsos , from the recent compilation by based on 105783 qsos of sdss dr7 @xcite .
however , the range of @xmath2 varies with @xmath15 , and the observed wavelength range of the spectrum .
therefore , in order to make the sample unbiased , firstly , we have considered a sdss spectral range from 4000 - 9000 which is a little narrower ( by about 100 ) than the actual one .
we then applied the following four selection filters . 1 .
we removed 773 broad absorption line ( bal ) qsos from our above sample to avoid any contamination in our analysis by bal features which has resulted in the removal of corresponding 931 strong absorbers .
2 . for all the quasars having @xmath18
, the emission line will fall above 9000 , which is our conservative upper limit on wavelength of sdss spectrum . as a result , sdss spectra for such sources will not allow any detection of strong doublet falling in the redshift range between 2.21024 up to @xmath15 . therefore to avoid this observational bias , we excluded all sources having @xmath19 from our sample , which resulted in the removal of 43 qsos having 52 strong mg ii absorbers .
3 . another filter was applied to avoid the observational bias which might result from the lower wavelength limit , viz 4000 , in the sdss spectra . in our analysis
we aim to see any correlation of luminosity with velocity offset up to about 0.4c .
however for 4000 considered as the conservative starting wavelength of our spectra , @xmath20 will be the minimum redshift , which allows us to detect absorber ( if any ) at least up to a velocity offset of @xmath21 .
therefore , we have removed 1461 sightlines with @xmath22 having 1544 strong absorbers in their spectra .
4 . after applying the above mentioned redshift cuts , we are left with the systems with @xmath23 . in these intermediate redshift systems ,
the @xmath2 value can be larger than @xmath24 , which in principle may give rise to a bias of higher @xmath2 with increasing @xmath15 .
hence we also remove all the absorbers with @xmath25 from the remaining sample which amounts to exclusion of 1523 absorbers along 1439 sightlines .
one should note that @xmath26 is chosen because if we keep @xmath2 value less or greater than this , then the sample is significantly reduced .
another motivation as will be clear in the coming sections , is that @xmath27 is an upper limit for the radiation ( dust ) driven outflows . finally ,
we are left with 6317 strong systems along 5682 qsos in the selected redshift range .
bolometric luminosities for the qsos in sdss dr7 are calculated in a recent study by @xcite .
we cross matched the qsos in our sample , with the catalogue described in shen s paper to obtain the bolometric luminosity .
we then removed two more absorbers whose qso luminosities were @xmath28 erg s@xmath29 .
our final bias free sample consists of 6315 strong systems with luminosity range @xmath30 erg s@xmath29 , with redshift range @xmath31 , and with the velocity offset range of @xmath32 . in figure 1
, the blue dashed line represents the distribution of strong mg ii absorbers in sdss - dr7 , compiled by , and the black solid line is the final sample selected for this study . of the samples of strong absorbers in sdss - dr7 .
the blue dashed line is for 10367 strong absorbers compiled by .
the black solid line represents the sample used in this work . ]
in order to test the dependence of @xmath2 on luminosity , we divide the sample in bins of bolometric luminosity .
most of the absorbers ( 5651 out of 6315 ) belong to qso sightlines having a luminosity range @xmath34 erg s@xmath29 .
we divide these 5651 systems into four bins of bolometric luminosity .
we also have two more bins , one for @xmath35 erg s@xmath29 , and another with @xmath36 erg s@xmath29 , the first having 27 systems and the second with 637 systems .
consider the case of the absorbers being distributed uniformly in the allowed range of @xmath16 ( which in turn is determined from the allowed range of @xmath2 ) , then the median value of @xmath2 should be independent of @xmath15 ( see appendix a for a proof ) .
hence , irrespective of the distribution of @xmath15 in a luminosity bin , the median value of beta should be same in all luminosity bins . to test this hypothesis
, we estimate the median , the lower 25 percentile and the upper 25 percentile of data in each of the above mentioned six luminosity bins .
we plot the median with circles , and the upper and lower percentiles as the end points of vertical dotted bars in figure 2 . of absorber .
the circles represent the median of data in a particular luminosity bin .
the upper and lower extreme of the dotted vertical lines give the location of upper and lower 25 percentile , respectively .
sizes of the luminosity bins are shown by the horizontal bars .
the solid , dotted and dash - dotted line represent the theoretical model discussed in 3.1 . ]
interestingly , we find that the median is not constant .
the data shows a correlation of @xmath2 with the @xmath1 .
the 5651 absorbers systems with @xmath37 erg s@xmath29 , which form the mainstay of the sample show a power law increase of @xmath2 with @xmath1 , with a slope of @xmath38 .
increase of median value of @xmath2 with the bolometric luminosity , proves that the distribution in each bin is not uniform random .
this fact is also hinted in the evolution of @xmath13 with @xmath16 @xcite . using the evolution in @xmath13 for our sample
, we evaluated the expected relation between the median value of @xmath2 and @xmath15 from equation ( [ app_med ] ) .
we then converted it to the corresponding relation between @xmath2 and luminosity by using the best fit relation between @xmath15 and luminosity , a characteristic of magnitude limited survey such as sdss .
we have shown this relation using a dashed curve in figure 2 .
although the dashed curve does show some evolution , but it is clear that it can not fully explain the observed @xmath39 correlation
. which physical processes can give rise to non - uniformity of absorber distribution ?
the evolution in @xmath13 has been attributed to the evolution in global star formation rate @xcite , although without any concrete evidence .
also , observations of intervening galaxies show a small covering fraction ( @xmath40 ) for strong absorbers ( @xmath41 ) @xcite . here
we explore an alternate based on outflows associated with qsos , which can give rise to non - uniformity of incidence of absorbers . as we explore in next section ,
the relation @xmath42 , is a natural consequence of qso radiation driven outflows .
radiation driven outflows have been invoked repeatedly in literature to explain the co - evolution of black hole and bulge , to explain the accretion disc winds ( e.g. * ? ? ?
* ) and galactic winds ( e.g. * ? ? ?
* ; * ? ? ?
we consider here the radiation driven outflows , where the photons scatter the dust grains and impart their momentum to dust .
the dust in turn is collisionally coupled to the gas , and the momentum is uniformly distributed to the dust and gas mixture . in this scenario , the motion of dust and gas mixture surrounding the qso is governed by the following equation , @xmath43 where @xmath44 is the mass of the black bole and @xmath45 is the dark matter halo potential .
@xmath46 is the integrated uv luminosity and for qsos where the main emission is in high frequency bands , luminosity over uv and euv bands is roughly half of the bolometric luminosity ( @xmath47)@xcite .
@xmath48 is the frequency averaged opacity for the scattering and absorption of uv photons by dust grains . for wavelength of photon @xmath49 m
, the @xmath48 for a dust and gas mixture ranges from 200 to as large as 1000 @xmath50g@xmath29 @xcite .
we take a value @xmath51 @xmath50g@xmath29 , which roughly serves as an average effective value of opacity .
we can integrate equation ( [ eq_diff ] ) to obtain the following expression for velocity @xmath52 where @xmath53 is the launching radius of the outflow . in the case of radiation pressure on dust grains ,
the opacity is generally quite high and hence the radiation force is many times larger than the gravity , therefore the gravitational force can be neglected . at a large distance
the velocity attains the following terminal value @xmath54 the base radius ( @xmath53 ) for launching these outflows is an important factor and it should be the minimum distance at which the dust grains can survive . studies on dust survival yield following relation between the sublimation radius of the dust grains and the luminosity of the agn @xcite , @xmath55 the value of @xmath56 is @xmath57 pc for graphite grains and @xmath58 pc for the silicate grains . substituting equation ( [ eq_mor ] ) into ( [ wind_easy ] ) , we obtain the following expression for wind terminal speed , @xmath59 we note that this mechanism has previously been discussed in the context of agn outflows by @xcite .
these authors also arrived at similar terminal speed for a radiation driven outflow .
we plot this scaling to compare with the observed correlation of @xmath2 and @xmath1 in figure 2 .
the dash - dotted , solid and dotted line in figure 2 correspond to @xmath60 pc respectively .
we find that this simple theoretical model fits the observed correlation pretty well , which indicates that the absorber systems are likely to be radiation(dust ) driven outflows .
one is then tempted to ask as to how these outflows fit in the unification schemes of agn .
we find that the launching radius of the outflows is the dust sublimation radius , which is also the inner radius of the dusty torus . inside the torus , the uv photons
are quickly reprocessed into ir .
although the ir photons can also drive outflows @xcite , however the speeds would not be large , as the ir to dust scattering cross section is more than an order of magnitude smaller than in uv .
one possible way to reconcile this is the following .
let us suppose that the outflows do not plough through the main body of the torus , but consist of material lifted from the outer surface of the torus . in that case , as the torus material is dilute and highly porous at its periphery , the uv photons can in principle travel a large distance without being attenuated and impart their momentum to gas and dust mixture lifted from the outer surface of the torus .
more specifically , in the picture presented in @xcite , the region which we are considering should take place between the bal envelope and the torus .
we note that , this scenario not only gives rise to large velocity outflows , but it may also account for the small fraction ( @xmath61 ) of the qsos which show these absorbers owing to the fact that the region allowed for the outflows ( periphery of the torus ) occupies a very small fraction of the viewing angle .
next we study the dependence of fractional number counts of absorbers as a function of qso luminosity .
we define the fractional number count as below , @xmath62 again , we limit our analysis to the spectral region with @xmath63 . from our sample , as described in 2 , we can easily estimate the number of absorbers found " in a given luminosity bin , having @xmath63 .
however to find the corresponding number of qsos searched in a bin " , we also need to count those qsos in the parent sample of sdss - dr7 from which the qsos with absorbers are selected .
we use the parent catalogue from @xcite of which the sample used in this work is a subset .
therefore , we estimate the number of qsos searched in a bin " by using non - bal qsos from shen et al ( 2012 ) catalogue , which satisfy the redshift criteria @xmath64 , to ensure the absence of any observational biases ( see 2 ) . we plot the fractional number count as a function of luminosity ( @xmath1 ) in figure 3 .
the values are shown by filled diamonds whose x - coordinate is the centre of each luminosity bin .
we also show the overall average of the sample using a horizontal line , whose value is @xmath65 .
we find that the fraction increases steeply with increasing qso luminosity and reaches a maximum roughly for @xmath66 erg s@xmath29 .
for @xmath67 erg s@xmath29 there is a mild decrease with luminosity , however this decrease is uncertain as in this bin , we have many apparently faint high redshift quasars , for which the signal to noise criterion removes large chunks of spectra and the corresponding absorbers ( d. lawther pvt . comm . ) .
we note that the fractional number of absorbers has a contribution from outflowing and intervening systems , which we can not separate here .
we would like to emphasize an important point in connection with our result .
there is a general consensus in the literature , which goes along the line that the absorbers which have @xmath68 ( @xmath69 km s@xmath29 ) , are associated with the qso and with @xmath2 higher than this represent the intervening media .
we stress here that this criterion is not adequate to denote the associated systems , and the true associated systems can also have @xmath70 , e.g. the qso driven high velocity outflows considered here .
to illustrate this , we plot in figure 4 , the ratio of absorbers with @xmath68 to the total number of absorbers in a particular luminosity bin as a function of bolometric luminosity .
one can clearly see that lower @xmath2 are possible for only lower luminosity , and vice versa .
firstly , the figure once again confirms that the velocity offset @xmath2 is correlated with luminosity , because low @xmath2 absorbers appear along the sightlines of low luminosity qsos .
secondly , this plot , in conjunction with the correlation of @xmath2 with @xmath1 , shows that the systems which are really ` _ associated _ ' with the qsos are spread all the way from @xmath71 to @xmath24 .
our results call for a study to separate out the truly associated ( outflowing ) systems and the intervening ones .
of course , one tedious way to do this is to locate the intervening galaxies in each quasar sightline , however yet another way can be through the detailed study of line shapes and features arising from outflows and intervening material .
we look forward to such a study in the future . to the total number of quasars in a particular luminosity bin
is plotted against the bolometric luminosity .
the horizontal bar represents the size of the luminosity bin .
] there is another implication of the observed dependence of fractional number count of absorbers on qso luminosity .
if one considers a sample of a particular type of qsos that covers a restricted luminosity range , then the relative number of absorbers may differ for different samples , and be different from the overall average .
if we consider the right side of figure 3 , corresponding to @xmath72 erg / s , there the fractional number counts are roughly double of the overall average value of @xmath73 .
we note that recent observations of @xmath74 blazars report an excess of absorbers relative to that in qsos .
we speculate here that this excess may also arise from the fact that the blazar sample is small , and it may be possible that it is biased towards higher luminosity , where the fractional number count is larger .
it is possible that if the analysis is repeated with a larger sample of blazars then the excess may fade away .
in fact , a similar conclusion has been reached for a sample of fsrqs and @xmath75 lobe and core dominated qsos where in both cases one finds only a mild excess @xcite . in this
regard we bring a recent paper by @xcite to the attention of the reader , regarding the excess seen towards grbs , where with a large sample of grbs the puzzle of incidence rate indeed disappears , and one does not find any excess . in summary
, we have found a correlation between the velocity offset of strong absorbers and the luminosity of qsos .
the velocity offset ( @xmath76 ) has been found to increase with the luminosity with a power law index @xmath3 .
we have found that radiation driven outflows from qsos can give rise to such a dependence of @xmath2 on @xmath1 .
these findings lead us to conclude that a significant fraction of strong absorbers ( even with @xmath77 @xmath6 ) along qso sightlines may be the agn driven outflows .
we are grateful to d. lawther for supplying the redshift path data .
we thank an anonymous referee for insightful comments .
from equation ( 1 ) , we know that @xmath2 mainly depends on the difference of @xmath78 and @xmath79 . here
we address the question whether or not the median value of @xmath2 which we have computed over its fixed range varying from @xmath80 , depends on the value of @xmath15 . for maximum value of @xmath81 , the lower limit of @xmath16 , using equation ( 1 )
is given by @xmath82 where @xmath83 is a constant whose value is 0.65 for @xmath84 .
the median value of absorber redshift ( @xmath85 ) is the solution of following equation , @xmath86 where @xmath13 is the number of absorbers per unit redshift .
if the absorbers are distributed uniformly and the quantity @xmath13 is constant , then from equation ( [ app_med ] ) we get , @xmath87 using equation ( 1 ) we find the corresponding median value of @xmath2 , which is @xmath88 independent of @xmath15 .
however , the observed @xmath89 evolves with luminosity ( and @xmath15 ) , which indicates that absorbers are not distributed uniformly . | we report a correlation between velocity offset ( @xmath0 ) of strong absorption systems and the bolometric luminosity ( @xmath1 ) of quasars in sdss - dr7 . we find that , @xmath2 shows a power law increase with @xmath1 , with a slope @xmath3 .
we find that such a relation of @xmath2 with @xmath1 is expected for outflows driven by scattering of black hole radiation by dust grains , and which are launched from the innermost dust survival radius .
our results indicate that a significant fraction of the strong absorbers , in the range of @xmath4 may be associated with the quasars themselves .
[ firstpage ] galaxies : active quasars : general quasars : absorption lines |
flow and transport are fundamental in many geo - engineering applications , including oil and gas recovery from hydrocarbon reservoirs , groundwater contamination and sustainable use of groundwater resources , storing greenhouse gases ( e.g. co@xmath1 ) or radioactive waste in the subsurface , or mining heat from geothermal reservoirs . in porous media , a single phase transport process is described by the equation @xmath2,\end{aligned}\ ] ] where @xmath3 is an open domain of @xmath4 , @xmath5 is the porosity ( void fraction ) of the rock , and @xmath6 is the symmetric dispersion tensor , @xmath7 is the unknown concentration of the contaminant , @xmath8 the darcy s velocity and @xmath9 the reaction and source term .
+ finite element , finite volume or the combination finite element - finite volume methods are mostly used for space discretization of the problem ( [ adr ] ) while explicit euler , semi implicit and fully implicit methods are usually used for time discretization @xcite . due to time step size constraints , fully implicit scheme is more popular for time discretization for quite a long time compared to explicit euler and semi implicit methods .
this method , however , need at each time step a solution of large systems of nonlinear equations
. this can be the bottleneck in computations . in recent years
, exponential integrators have become an attractive alternative in many evolutions equations @xcite .
in contrast to classical methods , they do not require the solution of large linear systems .
instead they make explicit use of the matrix exponential and related matrix functions .
the price to pay is the computing of the matrix exponential functions of the non diagonal matrices , which has revived interest and significance progresses nowadays @xcite . in this work ,
we combine a finite volume method with the first order exponential time differencing scheme of order 1 ( etd1 ) .
although both discretization techniques have been together used for solving evolutionary problems like ( [ adr ] ) ( see @xcite ) , a proper combination of rigorous proof of them has been lacking so far .
the paper is organised as follows . in section [ sec : semi ] we present the semi group formulation . in section [ sec : fvm ] and section [ sec : etd ] we present the space and time discretization .
we then state and discuss our main result in section [ sec : th1 ] and present the proof of the convergence theorem .
we end by presenting some simulations in section [ sec : simulation ] , these are applied both to a linear example where we can compute an exact solution as well as a more realistic model ( spe10 case @xcite ) .
let us start by presenting briefly the notation of the main function spaces and norms that we use in this paper .
we denote by @xmath10 the norm associated to the inner product @xmath11 of the hilbert space @xmath12 . for a banach space @xmath13
we denote by @xmath14 the norm of @xmath13 and @xmath15 the set of bounded linear mapping from @xmath13 to @xmath13 to @xmath16 . for the sake of simplicity , without loss of generality
we use the homogeneous dirichlet boundary condition and assume that the porosity @xmath5 is constant . during our simulation , we will also investigate the general boundary conditions and non constant porosity @xmath5 .
we assume that the darcy velocity @xmath8 is known , and satisfies the following mass conservation for incompressible fluids @xmath17 the model problem ( [ adr ] ) is reformulated as : find the concentration @xmath18 such that @xmath19 \\ x(\mathbf{x},0)=x_{0 } \quad \quad \quad \quad & \mathbf{x } \in \omega\\ x(\mathbf{x},t ) = 0 \quad \quad \quad \quad & ( \mathbf{x},t ) \in \partial \omega \times \left[0,t \right ] \end{cases } \end{aligned}\ ] ] where @xmath20 for well posedness of ( [ adr6 ] ) , we assume that @xmath6 is symmetric , @xmath21 and there exists a positive constant @xmath22 such that @xmath23 set @xmath24 , the bilinear form associated to the operator @xmath25 is given by @xmath26 according to grding s inequality ( see @xcite ) , there exists two positive constants @xmath27 and @xmath28 such that @xmath29 by adding @xmath30 in both side of the first equation of ( [ adr ] ) we have a new operator that we still call @xmath31 corresponding to the new bilinear form that we still call @xmath32 such that the following coercivity property holds @xmath33 we will still call the right hand side of the first equation of ( [ adr6 ] ) @xmath9 .
by green s formula we have @xmath34 where @xmath35 the domain of the operator @xmath31
. therefore the weak form of ( [ adr6 ] ) is to find the function @xmath36 such that @xmath37 \\ x(t)=x_{0}. \end{cases}\end{aligned}\ ] ] the @xmath38ellipticity ( [ ellip ] ) implies that @xmath39 is a sectorial on @xmath40 ( see @xcite ) i.e. there exists @xmath41 such that @xmath42 where @xmath43 .
then @xmath39 is the infinitesimal generator of bounded analytic semigroups @xmath44 on @xmath40 such that @xmath45 where @xmath46 denotes a path that surrounds the spectrum of @xmath47 .
the coercivity property in ( [ ellip ] ) implies also that the set of the real part of the spectrum of @xmath31 is non negative , which allows the definition of the fractional power of @xmath31 as : for any @xmath48 @xmath49 where @xmath50 is the gamma function of @xmath51 @xcite .
we denote by @xmath52 the norm of the space @xmath53 . for the nonlinear reaction term
we make the following assumption * [ lipschitz condition for @xmath9 ] * + [ alip ] the nonlinear @xmath9 is continuous and differentiable respect to @xmath7 with respect to the variable @xmath7 , with bounded differential
. therefore there exists a positive constant @xmath55 such that @xmath56 \\
\newline\\ \vert r(y)-r(z ) \vert \leq l \vert y - z\vert\;\;\;\forall y , z \in l^{2}(\omega ) , \end{array}\right.\end{aligned}\ ] ] where with a slight abuse of notation , @xmath9 denotes the nonlinear operator @xmath57 . by duhamel
s principle we may represent solutions of ( [ adr6 ] ) by the following integral equation @xmath58 . \ ] ] applying the contraction mapping principle in the topology of the banach space @xmath59,h^{1}(\omega))$ ] to the integral equation ( [ mild ] ) ( * ? ? ?
* theorem 3.3.3 ) or ( * ? ? ?
* theorem 6.3.1 ) ensure the existence and uniqueness of @xmath7 .
a cell centred finite volume methods for heterogeneous and anisotropic diffusion problems remains a challenging problem .
an active area of research is to make the approximation of the diffusion flux more efficient and simple as possible ( see @xcite for the references ) . the finite volume method is widely applied when the differential equations are in divergence form . to obtain a finite volume discretization ,
the domain @xmath3 is subdivided into subdomains @xmath60 being the corresponding set of indices , called control volumes or control domains such that the collection of all those subdomains forms a partition of @xmath3 .
the common feature of all finite volume methods is to integrate the equation over each control volume @xmath61 and apply gauss s divergence theorem to convert the volume integral to a surface integral . for our parabolic problem ( [ adr6 ] ) ,
finite volume methods differ in the way they approximate the diffusion flux @xmath62 .
two techniques are mostly used : the finite volume with two - point flux approximation ( tpfa ) ( see @xcite ) and the finite volume with multi - point flux approximations ( mpfa)(@xcite ) .
an advantage of the two - point approximation is that it provides monotonicity properties , under the form of a local maximum principle .
it is efficient and mostly used in industrial simulations . in this paper
we use the tpfa as developed in ( * ? ? ?
* page 815 ) .
the main drawback of tpfa is that it is applicable in the so called `` admissible mesh '' and not in general mesh .
[ admissiblemesh ] [ * an admissible mesh @xcite * ] an admissible mesh @xmath63 for problem ( [ adr6 ] ) with the full diffusion tensor @xmath6 is given by : * a set @xmath64 of control volumes such that @xmath65 with the corresponding local inner product induced by @xmath66 where @xmath67 * the corresponding set of center points @xmath68 such that 1 .
@xmath70 is the intersection of the straight lines perpendicular to the boundary of @xmath71 with respect to the inner product induced by @xmath66 .
let @xmath72 be the maximum mesh size of @xmath63 .
we denote by @xmath73 a dual delaunay triangulation of @xmath63 i.e. the delaunay triangulation where @xmath74 is the set of vertices .
let us illustre definition [ admissiblemesh ] to make it more understandable .
* in the case where the diffusion tensor @xmath6 is diagonal and @xmath3 is a rectangular or parallelepiped domain , any rectangular grid ( @xmath75 ) or parallelepiped grid ( @xmath76 ) is an admissible mesh .
the set @xmath77 is the set of centers of gravity of the rectangular grid or parallelepiped grid .
the inner product induced locally by @xmath66 is equivalent to the standard inner product corresponding to the euclidean norm @xmath78 .
this mesh will yield a 5-point scheme ( @xmath75 ) and 7-point scheme ( @xmath76 ) for our model problem ( [ adr6 ] ) . * if @xmath75 , for isotropic and heterogeneous media ( @xmath79 , @xmath80 being the identity matrix of dimension @xmath81 ) we can define a triangular admissible mesh @xmath82 to be a family of open triangular disjoint subsets of @xmath3 such that two triangles having a common edge have also two common vertices .
the angles of the triangles are assumed to be less than @xmath83 to allow the orthogonal bisectors to intersect inside each triangle , thus naturally defining the center point @xmath70 of the control volume @xmath71 .
the finite volume scheme defined on such mesh will yield a 4-point scheme for our model problem ( [ adr6 ] ) .
the inner product induced locally by @xmath66 is equivalence to the standard inner product corresponding to the euclidean norm @xmath78 . to make notation easier
, we will identify @xmath63 to @xmath84 , then to say @xmath85 we will say @xmath86 .
consider the modified model problem of ( [ adr6 ] ) where @xmath30 is added on both sides of the first equation of problem ( [ adr6 ] ) , @xmath27 is defined in ( [ coer ] ) .
consider an admissible mesh @xmath63 in the sense of definition [ admissiblemesh ] .
denote by @xmath87 the set of edges of control volume of @xmath88 the set of interior edges of control volume of @xmath89 , @xmath90 the approximation of @xmath7 at time @xmath91 at the center ( or at any point ) of the control volume @xmath92 and @xmath93 the approximation of @xmath7 at time @xmath91 at the center ( or at any point)of the edge @xmath94 . for a control volume @xmath92 ,
denote by @xmath95 the set of edges of @xmath96 , @xmath97 the lebesgue measure of the control volume @xmath92 . as in @xcite , integration over any control volume @xmath92 , using the divergence theorem to convert the integral over @xmath96 to a surface integral , finite differences for the diffusion flux approximation @xcite and the upwind technique for the advection flux approximation yields @xmath98 here
@xmath99 is the normal unit vector to @xmath100 outward to @xmath96 , @xmath101 is the lebesgue measure of the edge @xmath102 and @xmath103 the distance between the center of @xmath96 and the edge @xmath104 .
since the flux is continuous at the interface of two control volumes @xmath96 and @xmath105 ( denoted by @xmath106 ) we therefore have @xmath107 for @xmath108 , which yields @xmath109 with @xmath110 where @xmath111 is the distance between the center of @xmath96 and center of @xmath105 .
we will also denote by @xmath112 the distance @xmath111 or @xmath103 for @xmath113 or @xmath114 respectively .
notice that for @xmath115 , we can also write @xmath116 with @xmath117 the upwind term for advection flux @xmath118 is defined as @xmath119 we can write @xmath118 as @xmath120 where @xmath121 . the finite volume space discretization for the model problem ( [ adr6 ] ) is given by @xmath122 the scheme ( [ fvfd ] ) clearly indicates the affinity of the finite volume method to the finite difference method .
however , for the subsequent analysis it is more convenient to rewrite scheme ( [ fvfd ] ) in a discrete variational form .
multiplying the first equation of ( [ fvfd ] ) by arbitrary numbers @xmath123 and summing the results over all control volume in @xmath63 yields @xmath124 v_{i}= \underset{i \in \mathcal{t}}{\sum } \mathrm{mes}(i)\,r(x_{i}(t))v_{i},\\ \newline\\ r(x_{i}(t)):=r(\mathbf{x}_{i},t , x_{i}(t ) ) .
\end{array}\right.\end{aligned}\ ] ] let @xmath125 denote the space of continuous functions that are piecewise linear over the delaunay triangulation @xmath73 ( dual of @xmath63 ) , then the values @xmath90 and @xmath126 can be interpolated in @xmath127 .
there are unique functions @xmath128 such that @xmath129 and @xmath130 for all @xmath131 , where @xmath70 is a center of the control volume @xmath131 ( @xmath70 is also a vertex in @xmath73 ) .
denote by @xmath132 the bilinear form defined by @xmath133 and by @xmath134 the scalar product on @xmath135 defined by @xmath136 the corresponding norm is the discrete @xmath137 norm denoted by @xmath138 .
it is proved in @xcite that @xmath138 is equivalent to the @xmath137 norm @xmath139 in @xmath127 when the mesh @xmath63 is regular .
+ previous results allow us to write the following variational form of our finite volume scheme ( [ fvfd1 ] ) .
@xmath140,\\ \newline\\ x_{h}(0)=x_{h\,0},\\ \end{array}\right.\end{aligned}\ ] ] with @xmath141 consider the operator @xmath142 such that @xmath143 the semidiscrete solution in @xmath127 is then given by : find @xmath144 such that @xmath145 \\
\newline\\ x_{h}(0)=x_{0 h } \end{array}\right.\end{aligned}\ ] ] where @xmath146 is the orthogonal projection defined from @xmath147 to @xmath127 by @xmath148
in order to give the corresponding mild form of ( [ adrh ] ) and build the exponential integrator scheme let us define the discrete @xmath149 norm .
* [ discrete @xmath149 norm @xcite ] * + [ seminorm ] let @xmath63 be an admissible finite volume mesh in the sense of definition [ admissiblemesh ] .
let @xmath150 be the space of the functions constant in each control volume of @xmath63 . for @xmath151 , the discrete @xmath149 norm of @xmath152
is defined by @xmath153 where @xmath154 during our analysis we make the following assumption as in ( * ? ? ?
* page 815 ) .
[ meshassumption ] [ * regularity of @xmath6 , @xmath8 and the admissible mesh @xmath63 * ] + we assume that @xmath6 is bounded , @xmath155 and that there exists @xmath156 and @xmath157 such that @xmath158 assumption [ meshassumption ] allows the following @xmath159 ellipticity of @xmath132 .
[ coert ] under the regularity of the admissible mesh @xmath63 in assumption [ meshassumption ] , there exists a constant @xmath48 such that @xmath160 let @xmath161 and @xmath162 the bilinear forms defined in @xmath163 by @xmath164 using assumption [ meshassumption ] , mainly the regularity of @xmath63 and the fact that the coefficients of the diffusion tensor @xmath6 are bounded , there exists two constants @xmath165 and @xmath166 such that @xmath167 and @xmath168 so that @xmath169 where @xmath170 is defined in ( [ transn11 ] ) and ( [ transn1 ] ) .
using the fact that the transmissibility given in ( [ interf ] ) is symmetric , i.e. @xmath171 and reorganizing the summation , we therefore have @xmath172 let use some important results from @xcite . indeed as in @xcite reordering the summation over the set of edges yields @xmath173 where @xmath174 since @xmath175 we therefore have @xmath176 since we have assumed divergence - free flow .
then @xmath177 we also have @xmath178 therefore @xmath179 the following @xmath159 ellipticity of @xmath132 implies that @xmath180 is a sectorial on @xmath40 ( uniformly in @xmath72 ) i.e. there exists @xmath41 , such that @xmath181 where @xmath43 .
the discrete operator @xmath180 therefore is the infinitesimal generator of bounded analytic semigroup ( or exponential operator ) @xmath182 on @xmath127 such that @xmath183 where @xmath184 denotes a path that surrounds the spectrum of @xmath180 . as for the continuous case
, duhamel s principle implies that the solution of ( [ adrh ] ) is represented by the following integral equations ( mild form ) @xmath185.\end{aligned}\ ] ] for simplicity we consider a constant time - step @xmath186 . at time @xmath187 , $ ] the mild solution ( [ adrd ] ) is given by @xmath188 then , given the solution @xmath189 at the time @xmath190 , we can construct the corresponding solution at @xmath191 as @xmath192 note that the expression in ( [ exactnu ] ) is still an exact form of @xmath189 .
the idea behind exponential time differencing is to approximate @xmath193 by a suitable polynomial @xcite .
we consider the simplest case where @xmath193 is approximated by the constant @xmath194 and the corresponding scheme ( etd1 ) is given by @xmath195 where @xmath196 note that the etd1 scheme in ( [ etd ] ) can be rewritten as @xmath197 this new expression has the advantage that it is computationally more efficient as only one matrix exponential function needs to be evaluated at each step .
we assume that the unique mild solution @xmath7 of adr problem ( [ adr6 ] ) is the classical solution of ( [ adr6 ] ) i.e. @xmath7 is twice continuously differentiable with respect to @xmath198 and differentiable with respect to @xmath91 .
this assumption is necessary to achieve optimal orders of convergence in time and space .
[ t1 ] consider the mild solution @xmath7 of the adr ( [ adr6 ] ) and the numerical solution ( [ etd1 ] ) given by the etd1 scheme .
assume that assumption [ meshassumption ] is satisfied and the reaction function @xmath9 satisfies assumption [ alip ] .
set @xmath199 , assume that @xmath200 , then the following estimate holds : @xmath201 where @xmath202 the proof follows in section [ proofd ] , but we need some preparatory results first .
[ prop111 ] @xmath203 $ ] let @xmath89 be an admissible mesh in the sense of definition [ admissiblemesh ] and @xmath73 its dual delaunay triangulation ( @xmath204 are vertices of @xmath73 ) .
let @xmath205 defined by @xmath206 where @xmath207 is the nodal basis corresponding to @xmath208 in the sense of finite element method @xmath209 .
if @xmath210 , then there exists a positive constant @xmath211 such that the following estimate holds @xmath212 if @xmath213,c^{2}(\overline{\omega}))$ ] , then @xmath214.\end{aligned}\ ] ] see ( * ? ? ?
* section 3.4 , theorem 3.29 , page 139 or exercise 3.25 page 147 ) or ( * ? ? ?
* theorem 17.1 , page 132 ) . *
[ smoothing properties of the semi group@xcite ] * + [ prop1 ] let @xmath215 and @xmath216 , then there exists @xmath217 such that @xmath218 in addition , the following results hold @xmath219 @xmath220 @xmath221 where @xmath222 , @xmath223 .
[ lemmad1 ] let @xmath7 be the mild solution of ( [ adr6 ] ) given in ( [ mild ] ) .
let @xmath224,\;t_{1}<t_{2}$ ] .
the following estimates hold : * \(i ) if @xmath200 then @xmath225 for @xmath226 small enough .
* \(ii ) if @xmath200 and @xmath9 satisfies the lipschitz condition in ( [ lip ] ) then @xmath227 * part ( i)*. consider the difference @xmath228 so that @xmath229 we estimate each of the terms @xmath230 and @xmath231 . for @xmath230 , using proposition [ prop1 ] yields @xmath232 for the term @xmath233 , we have @xmath234 with @xmath235 we now estimate each term @xmath236 and @xmath237 . for @xmath238 @xmath239 for @xmath226 small enough , using proposition [ prop1 ] yields @xmath240 for @xmath241 ,
using the fact that the semigroup is bounded , we have @xmath242 hence @xmath243 combining previous estimations of @xmath230 and @xmath244 ends the proof of part ( i ) .
* proof of part ( ii ) * we consider again the difference in ( [ ddif ] ) .
the difference with the proof of part ( i ) comes from the estimation of @xmath245 .
this time we write @xmath246 if @xmath9 satisfies the lipschitz condition given in ( [ lip ] ) , then using the result in part ( i ) together with proposition [ prop1 ] yields @xmath247 for @xmath226 small enough
. we also have @xmath248 using the two transformations @xmath249 we find @xmath250 the estimate of @xmath245 ends the proof .
[ lemmad3 ] [ @xcite ] + let the sequence @xmath251 .
if the sequence of nonnegative numbers @xmath252 satisfies the inequality @xmath253 for @xmath254 and @xmath255 , then the following estimate holds : @xmath256 where the constant @xmath257 depends on @xmath258 . in the followng proof , @xmath259 and @xmath260
are positive constants .
we use the equivalence of the norms @xmath261 and @xmath138 in @xmath127 as the mesh @xmath63 is regular @xcite . using the triangle inequality yields @xmath262 let us estimate @xmath263 . integrating the shifted version ( by adding @xmath30 in both size ) of equation ( [ adr6 ] ) over each control volume @xmath264 and using the divergence theorem yields @xmath265 for @xmath266,\;\;i \in \mathcal{t}$ ] and @xmath102 using the same notation as in @xcite , let us set @xmath267 , \\ \newline\\
r_{i,\sigma}(t)= \dfrac{1}{\mathrm{mes}(\sigma)}\left [ q_{i,\sigma}x_{i,+}-\int_{\sigma } \mathbf{q } x(t)\cdot \mathbf{n}_{\sigma}\right ] , \\
\newline\\ \rho_{i}(t)=x(\mathbf{x}_{i},t)-\dfrac{1}{\mathrm{mes}(i)}\int_{i}x(\mathbf{x},t)d\mathbf{x},\\ \newline\\ \varrho_{i}(t)= \dfrac{1}{\mathrm{mes}(i)}\int_{i } r(\mathbf{x},t , x(\mathbf{x},t))d\mathbf{x}-r(\mathbf{x}_{i},t , x_{i}(t ) ) .
\end{array}\right.\end{aligned}\ ] ] assuming that the unique solution @xmath7 of ( [ adr6 ] ) is the regular , taylor expansion yields @xmath268 using assumption [ meshassumption ] , mainly the fact that the coefficients of the diffusion tensor @xmath6 and the darcy s velocity @xmath8 are regular , we can therefore used the following similar results in @xcite for general elliptic operators @xmath269 using the fact that @xmath9 is differentiable with respect to @xmath7 and @xmath198 , taylor expansion respectively yields @xmath270 where @xmath271 if the solution @xmath7 is differentiable with respect @xmath198 , one more taylor expansion yields @xmath272 using the lipschitz condition ( [ lip ] ) yields @xmath273 subtracting the first equation of ( [ fvfd ] ) from ( [ vdadr ] ) and using previous expressions yields @xmath274 with @xmath275 , \\
\newline\\ g_{i,\sigma}(t)=-\tau_{\sigma}(e_{j}(t)-e_{i}(t)),\quad \quad \sigma= i \vert j , \\
\newline\\ g_{i,\sigma}(t)=\tau_{\sigma}e_{i}(t),\quad \sigma \in\mathcal{e}_{i}\cap \partial \omega,\\ \newline\\ w_{i,\sigma}(t)=q_{i,\sigma}(x(\mathbf{x}_{\sigma,+},t)-x_{\sigma,+}(t ) ) , \end{array}\right.\end{aligned}\ ] ] and @xmath276 multipling equation ( [ derr1 ] ) by @xmath277 and summing for @xmath131 yields @xmath278 \\ \newline\\ = \underset{i \in \mathcal{t}}{\sum}e_{i}(t)\left [ \int_{i}\left(r(\mathbf{x},t , x(\mathbf{x},t))-r(\mathbf{x}_{i},t , x_{i}(t))\right)d\mathbf{x } \right ] \\
\newline\\ + \underset{i \in \mathcal{t}}{\sum}\left[c_{0 } \,\mathrm{mes}(i)\rho_{i}(t)e_{i}(t)-\underset{\sigma \in \mathcal{e}_{i}}{\sum}\mathrm{mes}(\sigma)e_{i}(t)(r_{i,\sigma}(t)+r_{i,\sigma}(t))-e_{i}(t)s_{i}(t)\right ] .
\end{array}\right.\end{aligned}\ ] ] let @xmath279 a piecewise constant function defined by @xmath280.\end{aligned}\ ] ] as in the proof of theorem [ coert ] , using the fact that @xmath169 where @xmath170 is defined in ( [ transn11 ] ) and ( [ transn1 ] ) yields @xmath281 where @xmath282 setting @xmath283 as in the proof of theorem [ coert ] , important results from @xcite yields @xmath284 using ( [ flux2 ] ) and ( [ react ] ) in the expression ( [ derr2 ] ) yields @xmath285 the continuity of the diffusion and advection flux at each interface yields @xmath286 set @xmath287 using the relation ( [ conti ] ) , the cauchy - schwarz inequality as in @xcite for stationary elliptic problems and reordering the summation over the edges yields @xmath288 using the fact that @xmath289 and relation ( [ flux1 ] ) yields @xmath290 for any constant @xmath217 , young s inequality yields @xmath291 using expression ( [ derr4 ] ) and ( [ yp ] ) in expression ( [ derr3 ] ) yields @xmath292 \leqslant ( c_{4}'/2 + 1)\vert e_{\mathcal{t}}(t)\vert_{0,h}^{2}+ c_{6}\,h^{2}\\ \newline\\ c_{6}=c_{6}(c_{1 } , c_{3},c_{3}',c_{4},c_{5 } ) . \end{array}\right.\end{aligned}\ ] ] bounding the left hand side of expression ( [ derr5 ] ) below yields @xmath293.\end{aligned}\ ] ] integrating both side of expression ( [ derr6 ] ) through interval @xmath294,\ ; 0\leq t\leq t $ ] yields @xmath295.\end{aligned}\ ] ] applying the discrete gronwall lemma [ lemmad3 ] yields @xmath296 then @xmath297 if @xmath199 , we have @xmath298 ( see @xcite ) , we therefore have @xmath299 let us estimate @xmath233 . from ( [ mmild ] ) and ( [ etd ] )
we have @xmath300 and @xmath301 the smoothing properties of the semigroup @xmath302 in proposition [ prop1 ] and the equivalence @xmath303 in @xmath127 yields @xmath304 and @xmath305 for @xmath306 $ ] , using proposition [ prop111 ] yields @xmath307 since @xmath308 , the equivalence @xmath309 and the uniform estimate of the term @xmath263 in @xmath310 $ ] yields @xmath311 using the lipschitz condition ( [ lip ] ) with ( [ errori ] ) and ( [ errorii ] ) yields @xmath312 we also have @xmath313 using lemma [ lemmad1 ] and the lipschitz condition ( [ lip ] ) yields @xmath314 using lipschitz condition ( [ lip ] ) , proposition [ prop111 ] and the equivalence @xmath315 in @xmath127 @xmath316 then @xmath317 combining estimates @xmath263 and @xmath233 yields @xmath318 applying the discrete gronwall lemma [ lemmad3 ] in ( [ lastest ] ) ends the proof .
the key point in ( etd1 ) scheme is the computing of the exponential function , the so called @xmath319 .
it is well known that a standard pad approximation for a matrix exponential is not an efficient method for large scale problems @xcite .
here we use the real fast leja points and the krylov subspace techniques to evaluate the action of the exponential matrix function @xmath320 on a vector @xmath321 , instead of computing the full exponential function @xmath322 as in a standard padapproximation .
the details of the krylov subspace technique are given in @xcite while more information about the real fast leja points can be found in @xcite .
the aims here is to analyse the convergence in space by comparing the numerical solution to the exact solution for a linear problem .
we consider here a linear reaction diffusion with exact solution given in @xcite .
the domain is defined as @xmath323 , @xmath324 , @xmath325 .
the initial time is given as @xmath326 .
this is necessary because the exact solution is not defined at the origin and at @xmath327 .
the dispersion tensor @xmath6 is diagonal and heterogeneous with coefficients given by @xmath328 a curved the velocity field ( figure [ fig01a ] ) is given explicitly by @xmath329 where @xmath330 and @xmath331 .
the finite volume mesh @xmath63 is the rectangular grid .
initial and boundary conditions are taken according to the exact solution @xcite , assuming an instantaneous release at a point @xmath332 .
we take a fixed time - step of @xmath333 .
we use here the real the fast lja to compute the action of the exponential function @xmath319 .
the final time here is @xmath334 .
the domain is @xmath336\times [ 0,l_{2}]\times [ 0 , l_{3}]$ ] , the finite volume mesh @xmath63 is the set of parallelepipeds with space discretization stepsize @xmath337 .
dimensions are @xmath338,@xmath339 .
we use for the absolute tolerance error @xmath340 and @xmath341 for the krylov subspace dimension . to have the velocity field we solve with our finite volume scheme the system @xmath342 where the dynamical viscosity is @xmath343 and the diagonal permeability tensor @xmath344 comes from @xcite .
we use the dirichlet boundary at @xmath345\right\rbrace \cup\left\lbrace \left\lbrace l_{1 } \right\rbrace \times\left\lbrace l_{2}\right \rbrace\times\left[0,l_{3}\right]\right \rbrace,\ ] ] and homogenous neumann boundary elsewhere such that @xmath346\\ \newline\\ 7997.92\;\text { psi}\quad\text{in}\quad\left\lbrace l_{1 } \right\rbrace \times\left\lbrace l_{2 } \right\rbrace\times \left[0,l_{3}\right ] \end{array}\right .
\newline\\ - k\,\nabla p ( \mathbf{x},t)\,\cdot\mathbf{n } & = & 0\quad\text{in}\quad\gamma_{n}= \partial \omega \backslash\gamma_{d}.\end{aligned}\ ] ] for the nonlinear advection diffusion reaction we use the following boundary conditions @xmath347\right\rbrace\times \left[0,t \right ] \newline\\ x&= & 1\quad\text{in}\quad \left\lbrace\left\lbrace l_{1 } \right\rbrace \times \left\lbrace l_{2 } \right\rbrace\times\left[0,l_{3}\right]\right\rbrace\times \left[0,t \right]\\ \newline\\ -(\mathbf{d}\nabla x)(\mathbf{x},t)\cdot\mathbf{n } & = & 0\quad\text{in}\quad\gamma_{n } \times \left[0,t \right ] \newline\\ x_{0}&=&0\quad \text{in}\quad\omega \quad(\text{initial solution})\end{aligned}\ ] ] where @xmath348 is the unit outward normal vector to @xmath349 .
the velocity field obtained in ( [ vel ] ) is piecewise constant in the mesh @xmath63 . for reaction function ,
we use the classical langmuir sorption isotherm given by @xmath350 , with @xmath351 .
this reaction function is obviously lipschitz .
the reference or exact solution is the numerical solution with the smaller time step size ( @xmath352 ) .
the final time here is @xmath353 seconds .
we take a uniform dispersion(diffusion ) tensor of @xmath354 .
figure [ fig01b ] shows a convergence of order @xmath355 for the time discretisation , which agrees with the predicted order in theorem [ t1 ] even we do nt have any information about the regularity of the exact solution .
in fact to obtain the optimal order of convergence in time , the exact solution should be differentiable respect to the time , which may be the case here since the initial solution is infinitely differentiable and we are dealing with parabolic problem .
m. a. christie and m. j. blunt , tenth spe comparative solution project : a comparison of upscaling techniques , 4(4 ) ( 2001 ) 308317 .
a. tambue , g. j.lord , and s. geiger .
an exponential integrator for advection - dominated reactive transport in heterogeneous porous media .
229(10):39573969 , 2010 . a. tambue , and s. geiger , and g. j. lord .
exponential time integrators for 3d reservoir simulation . , 2010 .
m. afif and b. amaziane convergence of finite volume schemes for a degenerate convection diffusion equation arising in flow in porous media . 191
( 2002 ) 52655286 .
m. hochbruck and a. ostermann . exponential integrators . , 209286 ( 2010 ) , doi:10.1017/s0962492910000048 .
r. eymard , t. gallouet , and r. herbin , finite volume methods , in : p. g. ciarlet , j. l. lions ( eds . ) , handbook of numerical analysis volume 7 , north - holland , amsterdam , 2000 , pp . 7131020 .
r. b. sidje .
expokit : a software package for computing matrix exponentials .
, 24(1):130156 , 1998 . b. amaziane and m. el ossmani convergence analysis of an approximation to miscible fluid flows in porous media by combining mixed finite element finite volume methods .
24(3)(2008 ) 799832 .
r. eymard , d. hilhorst , and m. vohrali a combined finite volume finite element scheme for the discretization of strongly nonlinear convection diffusion reaction problems on nonmatching grids .
doi : 10.1002/num.20449 , 2009 .
l. bergamaschi , m. caliari , a. martinez , and m. vianello .
comparing ljaand krylov approximations of large scale matrix exponentials .
, 3994:685692 , 2006 . s. m. cox and p. c. matthews .
exponential time differencing for stiff systems . , 176(2):430455 , 2002 .
a. tambue .
efficient numerical methods for porous media flow .
m. caliari , m. vianello , and l. bergamaschi . interpolating discrete advection diffusion propagators at lja sequences . , 172(1):7999 , 2004 . h. fujita and t. suzuki .
evolutions problems ( part1 ) . in : p.
g. ciarlet , j. l. lions ( eds . ) handbook of numerical analysis , vol ii , north - holland , amsterdam , pp .
789928 , 1991 .
a. pazy .
semigroups of linear operators and applications to partial differential equations . , 1983 .
d. henry .
geometric theory of semilinear parabolic equations . , 1981 .
r. eymard , and t. gallouet , and r. herbin .
a cell - centred finite - volume approximation for anisotropic diffusion operators on unstructured meshes in any space dimension .
, 26:326353 , 2006 . i. aavatsmark .
an introduction to multipoint flux approximations for quadrilateral grids .
, 6(3 - 4):405432 , 2002
. i. aavatsmark , t. barkve , .
be , and t. mannseth .
discretization on non - orthogonal , quadrilateral grids for inhomogeneous , anisotropic media .
, 127(1):214 , 1996 .
p. knabner and l. angermann , numerical methods for elliptic and parabolic partial differential equations solution , springer verlag , berlin , 2003 .
m. hochbruck , and a. ostermann .
exponential runge - kutta methods for parabolic problems .
, 53:323339 , 2005 . h. berland , b. skaflestad , and w. wright . a matlab package for exponential integrators .
, 33(1 ) , article no.4 , 2007 . | we analyze the finite volume spatial convergence combined with exponential time differencing of order one ( etd1 ) for temporal discretization of a non linear advection diffusion reaction equation ( adr ) modeling transport in porous media .
we derive the @xmath0 estimate under the assumption that the non linear reaction is globally lipschitz .
we illustrate the theoretical proof by some applications in 2d and 3d highly anisotropic and heterogeneous porous media including the spe10 case @xcite .
we compute the exponential of the non diagonal matrices arriving in the finite volume spatial discretization with the fast lja points technique and the krylov subspace technique .
finite volume , exponential integrator , porous media , advection diffusion reaction |
in the past few decades , there has been tremendous interests in the study of subluminal and superluminal light propagation @xcite .
the group velocity of light pulse can be reduced in bose - einstein condensate of sodium atom gas @xcite , and in hot gases @xcite , and even halted in vapor of rb atoms @xcite .
also , it can exceed the vacuum light speed , c , and can even become negative @xcite .
these experiments are based on the fact that electromagnetically induced transparency ( eit ) and electro magnetically induced absorption ( eia ) @xcite lead to a dispersion profile with a sharp positive or negative derivative @xcite .
anomalous dispersion was first studied in mechanical oscillators @xcite and was later applied by sommerfeld and brillouin @xcite to light propagation in absorptive opaque materials .
they showed theoretically that inside an absorption line , the dispersion can be anomalous , resulting in a group velocity faster than c , the vacuum speed of light .
such an anomalous velocity appears due to the wave nature of light @xcite .
talukder et al .
have shown femtosecond laser pulse propagation has switched from superluminal to subluminal velocities in an absorbing dye by changing the dye concentration @xcite .
shimizu et al . were able to control the light pulse speed with only a few cold atoms in a high - finesse microcavity by detuning the laser frequency from a cavity resonance frequency - locked to the atomic transition @xcite . in a series of papers @xcite , chiao and
coworkers showed theoretically that anomalous dispersion can occur inside a transparent material .
it was predicted that by using a gain doublet @xcite , it is possible to obtain a transparent anomalous dispersion region where the group velocity of light pulse exceeds c with almost no pulse distortion .
wang et al . used gain - assisted linear anomalous dispersion in atomic cesium gas , and the group velocity of a laser pulse in their experiment exceeded c and could even become negative , while the shape of the pulse was preserved @xcite .
the incoherent pumping fields can also play an important role in the controlling of the group velocity of light in dispersive media @xcite .
the double-@xmath0 setup is another scheme which provides a very rich spectrum of phenomena based on atomic coherence @xcite .
recently , the dispersion and the absorption properties of a weak probe field in simple multi - level quantum systems have been considered @xcite .
they showed that , by using a coherent and an indirect incoherent pump field the group velocity of light can be controlled .
there have been only few experimental and theoretical studies in which both free absorption superluminal and subluminal light propagation in a single system have been realized . in this article , the dispersion and the absorption properties of a weak probe field in a three - level v - type atomic system is investigated . by application of indirect incoherent pump fields
the effect of populating upper levels on optical properties of the atomic medium in presence of a strong coherent pump field is investigated .
it is shown that the slope of dispersion changes from positive to negative just by adjusting the intensity of the coherent or indirect incoherent pump fields .
it is demonstrated that the absorption free superluminal and subluminal light propagation appear in this system .
the model consists of a closed three - level v - type atomic system with ground level @xmath1 and two upper levels @xmath2 and @xmath3 as shown in fig.[fig : f1 ] . a strong coherent coupling field with rabi frequency @xmath4 drives transition @xmath5 , while a weak tunable probe field with rabi frequency @xmath6 is applied to transition @xmath7 . here
, @xmath8@xmath9 are the atomic dipole moments , and @xmath10 ( @xmath11 ) is the amplitude of the coupling ( probe ) field . by indirect incoherent pumps with rates @xmath12 and
@xmath13 the population of the ground level @xmath1 can be pumped to the exited levels @xmath2 and @xmath3 via some unspecified auxiliary states @xcite .
the spontaneous decay rates from states @xmath2 and @xmath3 to the ground level are denoted by @xmath14 and @xmath15 , respectively .
the density matrix equations of motion under the rotating wave approximation and in a rotating frame are : [ eq:1 ] @xmath16 where @xmath17 are the coherence decay rates given by @xmath18 @xmath19 and @xmath20 are the detunings of the coupling field and the probe field , respectively .
note that the interference due to the different spontaneous emission channels has been ignored .
the response of the atomic system to the applied fields is determined by the susceptibility @xmath21 , which is defined as @xcite @xmath22 where @xmath23 is the atom number density in the medium .
the real and imaginary parts of @xmath21 correspond to the dispersion and the absorption , respectively .
the so - called group index , @xmath24 , is also introduced where @xmath25 is speed of light in the vacuum and @xmath26 is the group velocity of the probe field @xcite . here
@xmath27 is the real part of @xmath21 .
, @xmath28 , @xmath29 ( solid ) , @xmath30(dashed ) , @xmath31 ( dot - dashed ) , and @xmath32 ( dotted).,width=302 ] eq.([eq:4 ] ) shows that , when @xmath27 is negligible the slope of dispersion has the major role in determination of the group velocity and the group index . in our notation the positive ( negative ) slope of dispersion corresponds to the increase ( reduction ) of the group index and the positive ( negative ) value of @xmath33 shows the attenuation ( amplification ) of the probe field .
it is apparent that a negative group index means a negative group velocity .
the steady state solutions of eqs.([eq:1 ] ) for the weak probe field approximation i.e. @xmath34 in the case of tuned coupling field ( @xmath35 ) are [ eq:5 ] @xmath36 for simplicity , from this point on , it is assumed that the rabi frequencies ( @xmath37 and @xmath38 ) are real numbers , the spontaneous decay rates are the same for all levels , i.e. @xmath39 , and other parameters are normalized to the spontaneous decay rate @xmath40 . in the following , using the derived expression for susceptibility @xmath21 , in terms of given system parameters , the response of the atomic system to the applied coherent and incoherent fields is studied .
it has to say that the main interest is in dispersion and absorption properties of the probe field around zero detuning , @xmath41 .
( red - solid lines ) , @xmath42 , @xmath43 , @xmath44 , and @xmath45 ( black - dashed lines ) ( a ) and @xmath46 ( red - solid line ) , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 ( black - dashed lines ) ( b ) via @xmath37 and @xmath12 .
the common parameters are @xmath53 , @xmath54 , @xmath55 , and @xmath41 .
, width=302 ] first the familiar effect of the coupling field in absence of the indirect incoherent pumps ( @xmath56 ) is mentioned . in this case due to application of the coupling field ,
the absorption line for probe field splits into two absorption lines and consequently the slope of dispersion changes from negative to positive . in fig.[fig : f2 ] the real ( a ) and imaginary ( b ) parts of the susceptibility are plotted versus the probe field detuning .
it shows that increasing the value of the rabi frequency , @xmath37 , leads to the appearance of a transparent region between the absorption lines around zero detuning , @xmath41 , which is an eit situation proportional to lossless subluminal light propagation . for the next step , while the strength of the second indirect incoherent pump is zero ( @xmath54 ) , the effects of driving transition @xmath57 by the coupling field and populating level @xmath3 is investigated . to get a rough view of this condition , it will be useful to draw some contour plots for @xmath58 and @xmath59 , via the coupling rabi frequency , @xmath37 , and the first indirect incoherent pump strength , @xmath12 , while @xmath13 and @xmath60 are zero . using eqs.([eq:5 ] ) and adjusting scale arbitrarily by letting @xmath61 and @xmath62 , the explicit expressions for @xmath58 and @xmath59 in terms of @xmath37 , @xmath12 , and @xmath13 at zero probe field detuning ( @xmath41 ) are @xmath63 and @xmath64 respectively , where @xmath65 , @xmath66 , @xmath67 , @xmath68 ( solid ) , @xmath31 ( dashed ) , @xmath32 ( dot - dashed ) , and @xmath69 ( dotted).,width=302 ] for @xmath54 ,
equations [ eq:6 ] and [ eq:7 ] are used to draw the plot contours of fig.[fig : f3]a and fig.[fig : f3]b , respectively . in fig.[fig : f3]a , the superluminal regions are colored and the subluminal region is left white .
in addition , the separating contours ( @xmath70 ) are plotted with red - solid lines . in this figure
the contour lines for ( @xmath71 , and @xmath45 are plotted in black - dashed style .
fig.[fig : f3]b illustrates the absorption property of the medium , @xmath72 . in this figure
the absorptive and amplifying regions are separated by the contour line of @xmath73 ( red - solid line ) and the darkness of the colored regions is a measure of their absorption or amplification magnitude for the probe field .
therefore a lighter color is used to illustrate the more transparent region .
finally , the contour lines of @xmath74 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 , are plotted in black - dashed style .
as fig.[fig : f3]a illustrates , there are two regions of anomalous dispersion and one region of normal dispersion . as determined by fig.[fig : f3]b , the first region of negative dispersion is due to the existence of an absorption line around zero probe field detuning and occurs where @xmath75 .
the solid lines of fig.[fig : f2 ] can be regarded as an example of this region .
the other region of negative dispersion happens where @xmath76 , while according to fig.[fig : f3]b it accompanies by gain . for the region of normal dispersion , as fig.[fig : f3]b demonstrates , the absorption property can take either a small positive or a large negative value , proportional to negligible attenuation or strong amplification of the probe field , respectively . ,
@xmath77 , @xmath78 , @xmath79 ( solid ) , @xmath80 ( dashed ) , and @xmath81 ( dot - dashed).,width=302 ] , @xmath66 , @xmath82 , @xmath54 ( solid ) , @xmath83 ( dashed ) , and @xmath84 ( dot - dashed).,width=302 ] , versus @xmath12 for the parameters @xmath67 , @xmath85 ( solid ) , @xmath31 ( dashed ) , @xmath32 ( dot - dashed ) , and @xmath69 ( dotted ) .
( b ) the group index against @xmath37 for the parameters @xmath67 , @xmath66 ( solid ) , @xmath86 ( dashed ) , @xmath81 ( dot - dashed ) , and @xmath87 ( dotted ) .
( c ) the group index versus @xmath12 while @xmath88 defined in eq .
( 8) for the parameters
@xmath89 ( solid ) , @xmath90 ( dashed ) , @xmath83 ( dot - dashed ) , and @xmath31 ( dotted ) .
other common parameters are @xmath53 and @xmath91.,width=302 ] fig.[fig : f4 ] and fig.[fig : f5 ] are presented to clarify how positive and negative dispersions can occur in the gain region of @xmath92 and @xmath12 plane . by the indirect incoherent pump of strength @xmath93 ( see eq.[eq:5c ] ) , the inversion of population between levels @xmath1 and @xmath3 is established .
in this situation the system shows amplification for the probe field . then there exists a gain deep with positive slope of dispersion around zero detuning ( see the red - solid lines of fig.[fig : f4 ] ) . by applying the coupling field level @xmath94
is split into two symmetric dressed levels @xmath95 and @xmath96 with frequency differences of @xmath97 from the original level .
when the coupling field is weak , there still seems to be one gain deep and the dressed levels are indistinguishable . by strengthening the coupling field ,
the gain deep is transformed to two gain deeps ( see fig.[fig : f4 ] ) .
this is proportional to changing the normal slope of dispersion to anomalous ( negative ) dispersion which leads to exceeding the group velocity from @xmath25 ( speed of light in vacuum ) or even may cause its becoming negative .
when the dressed levels are partly distinguishable , system may still exhibits a considerable gain around zero detuning , which corresponds to existence of noise in the medium for a probe pulse with a middle frequency of @xmath98 . by choosing a larger rabi frequency for the coupling field ,
the gain becomes lesser and the anomalous region becomes wider .
then to achieve a transparent medium with anomalous dispersion , i.e. superluminal light propagation , a strong enough coupling field should be applied while the population is inverted for the probe field transition @xcite .
if @xmath54 , as fig.[fig : f3]a shows , @xmath99 is the sufficient condition to switch the probe pulse propagation from subluminal to superluminal by application of the indirect incoherent pump of strength @xmath12 . for @xmath100
, @xmath101 is the threshold value at which the slope of dispersion changes from positive to negative . on the other hand ,
the steepest anomalous slope of dispersion occurs when @xmath102 and a strong enough indirect incoherent pump , populates level @xmath103 .
fig.[fig : f5 ] is presented to see how an increase in @xmath12 causes the gain deeps to become more distinguishable and consequently the anomalous slope of dispersion become steeper . in fig.[fig : f5]a for @xmath79 ( solid line ) the slope of dispersion around zero probe field detuning is positive . with increasing the indirect incoherent pump rate to @xmath66 ( dashed lines )
, the slope of dispersion becomes zero and it is negative for @xmath104 ( dash dotted lines ) . in fig.[fig : f5]b , the probe field absorption curves are plotted .
it shows that the subluminal and superluminal light propagation around zero probe detuning is accompanied by a considerable gain .
it is time to investigate the optical response of the medium to populating level @xmath105 while the transition @xmath106 is driven by the coherent field , and the population between levels @xmath107 and @xmath103 is inverted due to the indirect incoherent pump with a rate @xmath108 . in fig.[fig : f6 ] , the real ( a ) and imaginary ( b ) parts of @xmath21 are plotted versus probe field detuning .
the common parameters are @xmath109 , and @xmath110 .
the first plots ( solid lines ) , corresponding to @xmath54 , are presented for comparison only .
they show that the slope of dispersion is almost zero around zero probe field detuning and there is a considerable amount of gain due to the partly combined gain deeps . by populating level @xmath111 with rate @xmath112 , ( dashed lines )
, the slope of dispersion becomes negative around @xmath113 .
the absorption curve varies according to the probe field detuning , so that the maximum reduction in gain occurs around zero probe field detuning , while beyond @xmath114 , the gain increases slightly . by increasing the value of @xmath13 to @xmath115 ( dashed - dotted lines ) the region of anomalous
dispersion becomes wider and its negative slope increases in absolute value .
the absorption curve undergoes variations similar to those of the previous case ( @xmath112 ) .
the value of the gain for the latter case ( @xmath116 ) vanishes at @xmath117 .
the proper value of @xmath13 resulting in zero gain around @xmath113 can be obtained by : @xmath118 fig.[fig : f7 ] displays the effect of incoherent and coherent pump fields on the group index .
fig.[fig : f7]a shows the group index versus indirect incoherent pump rate @xmath12 for @xmath85 ( solid ) , @xmath31 ( dashed ) , @xmath32 ( dot - dashed ) , @xmath69 ( dotted ) , while @xmath67 .
fig.[fig : f7]b displays group index versus coherent pump field , @xmath37 , for @xmath66 ( solid ) , @xmath86 ( dashed ) , @xmath81 ( dash - dotted ) , and @xmath87 ( dotted ) , while @xmath54 . in fig.[fig
: f7]c , the group index is plotted as a function of @xmath12 for @xmath29 ( solid ) , @xmath119 ( dashed ) , @xmath83 ( dot - dashed ) , and @xmath31 ( dotted ) while the second indirect incoherent pump has rate @xmath120 defined in eq.([eq:8 ] ) .
it shows that even for @xmath121 the probe pulse can propagate superluminally while the medium is transparent , if proper values for @xmath12 and @xmath13 are chosen .
in conclusion , we have controlled the dispersion and the absorption of a weak probe field in a three - level v - type atomic system . by application of indirect incoherent pump fields
the effect of populating upper levels on optical properties of the atomic medium in presence of a strong coherent pump field has been investigated and it has been found that linear positive or negative transparent dispersion could occur between the doublet absorption or gain lines , respectively . then the absorption free superluminal light propagation has been established in this system .
s. e. harris , phys . today 50 ( 1997 ) 36 .
m. xiao , y. q. li , s. z. jin , j. gea - banacloche , phys .
74 ( 1995 ) 666 .
lord rayleigh , phil . mag .
xlviii , 151 ( 1899 ) .
l. brillouim , wave propagation and group velocity ( academic press , new york , 1960 ) .
r. y. chiao , phys .
rev . a 48 ( 1993 ) r34 .
e. l. bolda , r. y. chiao , and j. c. garrisson , phys .
a 48 ( 1993 ) 3890 .
a. m. steinberg and r. y. chiao , phys .
a 49 ( 1994 ) 2071 .
m. w. mitchell and r. y. chiao , am . j. phys .
66 ( 1998 ) 14 .
e. l. bolda , j. c. garrison , and r. y. chiao , phys . rev .
a 49 ( 1994 ) 2938 .
r. y. chiao and a. steinberg , in progress in optics , ed .
e. wolf ( elsevier , amsterdam , 1997 ) , p. 345 .
a. m. steinberg , p. g. kwiat , and r. y. chiao , phys .
71 ( 1993 ) 708 . m. fleischhauer , c. h. keitel , m. o. scully , c. su , opt . commu .
87 ( 1992 ) 109 . m. mahmoudi , m. sahrai , h. tajalli , phys . lett . a 357 ( 2006 ) 66
m. mahmoudi , j. evers , phys .
a 74 ( 2006 ) 063827 .
j. qian , h. zhang , j. opt .
b 21 ( 2004 )
1364 . m. mahmoudi , s.worya rabiei , l. ebrahimi zohravi , and m. sahrai , opt commun . 281 ( 2008 ) 4681 .
d. bullock , j. evers , c. h. keitel , phys .
lett . a 307 ( 2003 )
m. mahmoudi , s. w. rabiei , l. safari , m. sahrai , laser phys .
( 2009 ) 1428 - 1433 .
m. o. scully , m. s. zubairy , quantum optic , cambridge university press , london , 1997 . g. s. agarwal , t. nath dey , s. menon , phys .
a 64 ( 2001 ) 053809 . | dispersion and absorption properties of a weak probe field in a three - level v - type atomic system is studied . by application of indirect incoherent pump fields
the effect of populating upper levels on optical properties of the atomic medium in presence of a strong coherent pump field is investigated .
it is shown that the slope of dispersion switches from positive to negative just by changing the intensity of the coherent or indirect incoherent pump fields .
it is demonstrated that the absorption - free superluminal and subluminal light propagation appear in this system . |
the problem of determining the appearance of a fixed _ pattern _ in long sequences of observation is relevant in many scientific problems .
for example , in the area of computer network security , the detection of intrusions , which become increasingly frequent , is very important .
intrusion detection is primarily concerned with the detection of illegal activities and acquisitions of privileges that can not be detected by information flow and access control models .
there are several approaches to intrusion detection , but recently this subject has been studied in relation to pattern matching ( see @xcite ) .
this leads to the study of the construction of particular words avoiding a given pattern in an alphabet @xmath2 .
the present paper aims to be a contribution in this direction .
let @xmath3 be the set of binary words @xmath4 such that @xmath5 , for any @xmath6 , @xmath7 and @xmath8 corresponding to the number of 0 s and 1 s in the word @xmath4 , respectively .
in this paper we study the construction of the subset @xmath9 } \subset f$ ] of binary words excluding a given pattern @xmath10 , that is a word @xmath11}$ ] if and only if it does not contain a sequence of consecutive indices @xmath12 such that @xmath13 . if we consider the set of binary words without any restriction , the defined language is regular and we can refer to using classical results ( see , e.g. , @xcite ) .
when the restriction to words with no more 0 s than 1 s is valid , the language @xmath14}$ ] is not a regular one and it becomes more difficult to deal with .
for example , in order to generate the language @xmath14}$ ] for each forbidden pattern @xmath15 an `` ad hoc '' grammar should be defined .
our aim is to determine a constructive algorithm suggesting a more unified approach which makes it possible to generate all binary words in the class @xmath14}$ ] . in this paper
we show how to obtain all binary words belonging to @xmath16 and avoiding the pattern @xmath17 , for any fixed @xmath1 .
we @xcite introduced an algorithm for the construction of all binary words in @xmath16 having a fixed number of 1 s and excluding those containing the forbidden pattern @xmath18 , for any fixed @xmath19 .
that algorithm generates all the words in @xmath16 then eliminates those containing the forbidden pattern . basically , the construction marks in an appropriate way the forbidden patterns in the words and generates @xmath20 copies of each word having @xmath21 forbidden patterns such that the @xmath22 instances containing an odd number of marked forbidden pattern are annihilated by the other @xmath22 instances containing an even number of marked forbidden patterns .
for example , the words @xmath23 and @xmath24 , containing two copies of the forbidden pattern @xmath25 , ( the marked forbidden patterns are over - lined ) are eliminated by the words @xmath26 and @xmath27 , respectively .
this is possible since no prefix of @xmath28 is also a suffix of @xmath15 , that is the forbidden patterns do not overlap and so they are univocally identified inside the words .
then , the algorithm in @xcite can not be used to generate the words in @xmath14}$ ] when @xmath17 since the forbidden patterns may overlap inside the words . for example , in @xmath29 there are two overlapping copies of the forbidden pattern @xmath30 .
so , we propose a new algorithm that generates right the words in @xmath16 avoiding the forbidden pattern @xmath17 , for any fixed @xmath19 .
the paper is organized as follows . in section [ sec :
basic ] we give some basic definitions and notation . in particular , we recall how every binary word can be represented as a path on the cartesian plane . in section [ sec : algo ]
we give a construction , according to the number of 1 s , for the set of binary words excluding the pattern @xmath17 , for any fixed @xmath19 , and such that the number of 0 s in each word is inferior to or the same as the number of 1 s . in section [ sec :
proof ] we prove that the construction given in section [ sec : algo ] allows us to obtain an exhaustive and univocal generation of such binary words having @xmath31 1 s .
let @xmath3 be the set of binary words @xmath4 such that @xmath5 , for any @xmath6 , @xmath7 and @xmath8 corresponding to the number of 0 s and 1 s in the word @xmath4 , respectively .
in this paper we study the construction of the subset @xmath9 } \subset f$ ] of binary words excluding a given pattern @xmath32 , for any fixed @xmath19 . given @xmath33 the _ length _ of @xmath6 , we denote by @xmath34 , ( @xmath35 ) , the word with length @xmath36 obtained by linking @xmath4 to itself @xmath37 times , that is @xmath38 and @xmath39 , @xmath40 being the empty word .
each word @xmath6 can be naturally represented as a path on the cartesian plane by associating a _
rise _ ( or _ up _ ) _ step _ , defined by ( 1,1 ) and indicated by @xmath41 , with each bit 1 in @xmath4 and a _ fall _ ( or _ down _ ) _ step _ , defined by ( 1,-1 ) and indicated by @xmath42 , with each bit 0 in @xmath4 . for example , the word @xmath43 is represented by the path @xmath44 ( see figure [ fig : path ] ) .
an _ up - down _ step is the sequence @xmath45 . from now on , we refer interchangeably to words or their graphical representation on the cartesian plane , that is paths .
so by @xmath9}$ ] we denote both the set of pattern @xmath15 avoiding binary words and the set of corresponding paths .
in the rest of this paper , a path is defined as : * _ primitive _ if it begins and ends at ordinate 0 and remains strictly above the @xmath41-axis , * _ positive _ if it begins at ordinate 0 and remains above or on the @xmath41-axis , * _ negative _ if it begins and ends at ordinate 0 and remains below or on the @xmath41-axis ( remark that a negative path in @xmath16 necessarily ends at ordinate 0 ) , * _ strongly negative _ if it begins and ends at ordinate -1 and remains below or on the line @xmath46 , * _ underground _ if it ends with a negative suffix .
the _ complement _ of a path @xmath47 is the path @xmath48 obtained from @xmath47 by switching rise and fall steps .
in this section we show the constructive algorithm to generate the set @xmath9}$ ] , @xmath49 for any fixed @xmath1 , according to the number of rise steps , or equivalently to the number of 1 s . given a path @xmath11}$ ] with @xmath31 rise steps ,
we generate a given number of paths in @xmath9}$ ] with @xmath50 rise steps , @xmath51 , by means of constructive rules .
the number and the shape of the generated paths depend on the ordinate @xmath52 of the endpoint of @xmath4 and on its suffix . with regard to @xmath52 , we can point out three cases : @xmath53 , @xmath54 and @xmath55 , while as for the suffix we consider whether it is equal to @xmath56 or not .
when @xmath53 , we must pay attention also to the case in which @xmath4 is an underground path ending with the pattern @xmath57 . as we will show further on , for each @xmath11}$ ] such that @xmath53 or @xmath55 , the generating algorithm produces two or more positive paths and one underground path with @xmath50 rise steps , @xmath51 , while , when @xmath54 , it produces only one positive path with @xmath50 rise steps .
let us denote by @xmath58 a path with endpoint at ordinate @xmath52 .
the generating algorithm of the class @xmath9}$ ] with @xmath59 , for any fixed @xmath19 , is described in the following sections .
the constructive rules related to the special cases in which the suffix of @xmath4 is @xmath56 or @xmath57 are described in sections [ sec : positivesuffix ] and [ sec : negativesuffix ] , respectively , while in section [ sec : simplecase ] we examine all the other simple cases . the starting point of the algorithm is the empty word @xmath40 . in this section
we describe the constructive rules to be applied when the suffix of @xmath4 is neither @xmath56 nor @xmath57 .
we point out three cases for the ordinate @xmath52 of the endpoint of @xmath4 : @xmath53 , @xmath54 and @xmath55 .
: : a path @xmath11}$ ] , with @xmath31 rise steps and such that its endpoint has ordinate 0 , generates , for any @xmath37 , @xmath51 , three paths with @xmath50 rise steps : a path ending at ordinate 1 by adding to @xmath4 a rise step and a sequence of @xmath60 up - down steps ; a path ending at ordinate 0 by adding to @xmath4 a rise step , a sequence of @xmath60 up - down steps and a fall step , and an underground path obtained by the one generated in the previous step and mirroring on @xmath41-axis its rightmost primitive suffix
. + figure [ fig : zero1 ] shows the above described operations ; the number above the right arrow corresponds to the value of @xmath37 . both in this figure and in the following ones we consider @xmath61 , that is @xmath62
. + therefore @xmath63 @xmath54 .
: : a path @xmath11}$ ] , with @xmath31 rise steps and such that its endpoint has ordinate 1 , generates , for any @xmath37 , a path with @xmath50 rise steps with endpoint at ordinate 2 obtained by adding to @xmath4 a rise step and a sequence of @xmath60 up - down steps ( see figure [ fig : uno1 ] ) . + therefore @xmath64 @xmath55 .
: : a path @xmath11}$ ] , with @xmath31 rise steps and such that its endpoint has ordinate @xmath52 , @xmath55 , generates , for any @xmath37 , @xmath65 paths with @xmath50 rise steps : a path ending at ordinate @xmath66 by adding to @xmath4 a rise step and a sequence of @xmath60 up - down steps ; @xmath67 paths ending at ordinate @xmath68 , respectively , by adding to @xmath4 a rise step , a sequence of @xmath69 , @xmath70 , fall steps and a sequence of @xmath60 up - down steps ; a path ending at ordinate 0 by adding to @xmath4 a rise step , a sequence of @xmath52 fall steps , a sequence of @xmath60 up - down steps and a fall step , and an underground path which will be described in section [ sec : under ] .
figure [ fig : due1 ] shows the above described operations .
+ therefore + @xmath71 at this point it is clear that : 1 .
when the path @xmath4 ends with the suffix @xmath56 the paths obtained by means of the constructions ( [ alfa ] ) , ( [ beta ] ) and ( [ gamma ] ) contain the forbidden pattern @xmath72 .
so , we will act as described in section [ sec : positivesuffix ] ; 2 . when @xmath4 is an underground path ending with the pattern @xmath57 , some paths generated by means of the above constructions might contain the forbidden pattern @xmath72 .
so , we will follow a different procedure described in section [ sec : negativesuffix ] .
even when the path @xmath4 ends with the suffix @xmath56 , the number and the shape of the generated paths depend on the ordinate @xmath52 of the endpoint of @xmath4 .
let @xmath73 be the suffix of @xmath4 .
: : a path @xmath11}$ ] , with @xmath31 rise steps and such that its endpoint has ordinate 0 , generates , for any @xmath37 , @xmath74 , three paths with @xmath50 rise steps ( see figure [ fig : zero2 ] ) : a path ending at ordinate 1 , by inserting a sequence of @xmath60 up - down steps and a rise step on the left of @xmath75 ; a path ending at ordinate 0 , by inserting a sequence of @xmath60 up - down steps and a rise step on the left of @xmath75 and adding a fall step at the end of @xmath4 , and an underground path , obtained by mirroring on @xmath41-axis the rightmost primitive suffix of the path generated at the previous step . therefore + @xmath76 @xmath54 .
: : a path @xmath11}$ ] , with @xmath31 rise steps and such that its endpoint has ordinate 1 , generates , for any @xmath37 , a path with @xmath50 rise steps with endpoint at ordinate 2 , obtained by inserting a sequence of @xmath60 up - down steps and a rise step on the left of the suffix @xmath75 ( see figure [ fig : uno2 ] )
. therefore + @xmath77 @xmath55 .
: : a path @xmath11}$ ] , with @xmath31 rise steps and such that its endpoint has ordinate @xmath52 , @xmath55 , generates , for any @xmath37 , @xmath65 paths with @xmath50 rise steps ( see figure [ fig : due2 ] ) : a path ending at ordinate @xmath66 , by inserting a sequence of @xmath60 up - down steps and a rise step on the left of the suffix @xmath75 ; @xmath67 paths ending at ordinate @xmath68 , respectively , by inserting a sequence of @xmath60 up - down steps , a rise step and a sequence of @xmath69 , @xmath78 , fall steps on the left of @xmath75 ; a path ending at ordinate 0 , by inserting a sequence of @xmath60 up - down steps , a rise step and a sequence of @xmath52 fall steps on the left of @xmath75 , and then adding a fall step at the end of @xmath4 , and an underground path which will be described in section [ sec : under ] .
therefore + @xmath79 the paths @xmath80}$ ] ending on the @xmath41-axis with the sequence @xmath57 have the following shape @xmath81 where @xmath82 is a path ending on the @xmath41-axis and @xmath83 is either the empty path @xmath40 or is a strongly negative path .
the constructions applied to paths ending at ordinate 0 described in ( [ alfa ] ) ( see figure [ fig : zero1 ] ) can be used even for the paths ending with the sequence @xmath57 , when @xmath84 , or to generate the paths ending at ordinate 1 or on the @xmath41-axis with a positive suffix , when @xmath85 . nevertheless , when @xmath85 , by applying the construction , we obtain an underground path which contains the forbidden pattern @xmath86 . therefore if the path ends with the sequence @xmath57 and @xmath85 , in order to generate the underground path we proceed as follows .
two cases must be taken into consideration .
1 ) @xmath82 does not end with a peak @xmath45 .
: : the underground path generated from @xmath87 is obtained by adding the path @xmath88 to @xmath89 , mirroring on @xmath41-axis the rightmost suffix @xmath90 of @xmath91 and shifting the sequence @xmath56 between @xmath82 and the sub - path @xmath92 .
+ so the path @xmath87 generates the underground path @xmath93 ( see figure [ fig : casoa ] ) .
it should be noticed that this construction applies to @xmath4 even if @xmath94 .
2 ) @xmath82 ends with a peak @xmath45 . : : when the path @xmath82 ends with a peak @xmath45 , that is @xmath95 , the insertion of the sequence @xmath56 between @xmath82 and the sequence @xmath96 produces the forbidden pattern @xmath86 .
let us consider the following subcases : @xmath97 and @xmath98 .
+ 2.1 ) @xmath99 . ; ; the underground path is obtained by performing on @xmath100 the following operations : shifting the rightmost peak @xmath45 of @xmath82 to the right of the sub - path @xmath92 , mirroring on @xmath41-axis the sequence @xmath101 and adding to such path the steps @xmath102 .
+ so , when @xmath85 , the underground path with negative suffix generated by @xmath103 is @xmath104 ( see figure [ fig : casob1 ] ) .
2.2 ) @xmath105 .
; ; in this case , the underground path obtained by means of the construction described in 2.1 ) is @xmath106 and it contains the forbidden pattern @xmath86 if @xmath107 ends with the sequence @xmath90 or with the sequence @xmath108 , where @xmath109 is a not empty strongly negative path .
let us take the longest suffix of @xmath110 into account so that @xmath111 , where @xmath112 and @xmath113 is the empty path or is a strongly negative path .
every sequence @xmath114 , @xmath115 , will be changed into @xmath116 in the following way : + 2.2.1 ) : : if @xmath47 is a path that does not end with a peak @xmath117 , then @xmath118 and the underground path generated by @xmath89 is @xmath119 ( see figure [ fig : casob2i ] ) ; 2.2.2 ) : : if @xmath47 ends with a peak @xmath45 , that is @xmath120 , then @xmath121 and the underground path generated by @xmath89 is @xmath122 ( see figure [ fig : casob2ii ] ) .
now let us describe how to obtain the underground path generated by @xmath123 .
for any @xmath37 , @xmath51 , let @xmath124 be the path obtained from @xmath58 and ending on the @xmath41-axis with a positive suffix , @xmath47 is the rightmost suffix in @xmath125 which is primitive . if the path @xmath48 does not contain the forbidden pattern @xmath15 , the underground path generated by @xmath58 is @xmath126 .
if the path @xmath48 contains the forbidden pattern @xmath15 , we must apply a _ swap _
operation @xmath127 in order to obtain a path @xmath128 avoiding the forbidden pattern .
the underground path generated by @xmath58 is @xmath129 . before describing the @xmath127 operation on @xmath48 ,
let us consider the following proposition .
[ prop1]let @xmath130}$ ] a primitive path ; @xmath131 contains the forbidden pattern @xmath86 if and only if @xmath82 contains the pattern @xmath132 . from proposition [ prop1 ]
it follows that , if @xmath48 contains the forbidden pattern @xmath15 , then it is preceded and followed by at least a rise step .
operation @xmath127 must generate a path @xmath133 avoiding the forbidden pattern @xmath86 and such that @xmath134}$ ] ; in this way @xmath133 is not the complement of any path in @xmath9}$ ] .
the path @xmath135 is obtained in the following way : * consider the straight line @xmath136 from the beginning of the pattern @xmath137 and let @xmath138 be the rightmost point in which @xmath136 intersects @xmath48 on the left of @xmath15 such that @xmath138 is preceded by at least two fall steps ; * let @xmath139 , @xmath140 , the subsequence on the right of @xmath138 , followed by at least a fall step ; * _ swap _ the initial subsequence @xmath141 of @xmath15 and @xmath142 .
let us remark that @xmath142 can not be equal to @xmath56 as @xmath47 does not contain the forbidden pattern @xmath72 ( see figure [ fig : swap].a ) ) .
when @xmath143 , that is @xmath142 is the empty word , we simply insert @xmath144 into @xmath138 ( see figure [ fig : swap].b ) ) .
operation @xmath127 is applied to each forbidden pattern in @xmath48 .
[ prop2 ] let @xmath145 , then @xmath146}$ ] .
* proof . *
the @xmath127 operation transforms the subsequence @xmath147 , ( @xmath148 ) , of @xmath48 into the subsequence @xmath149 of @xmath133 .
the complement of @xmath150 is @xmath151 so @xmath152 contains the forbidden pattern @xmath72 . @xmath153
[ prop3 ] let @xmath154}$ ] a primitive path such that @xmath155}$ ] .
then there exists a path @xmath156}$ ] such that @xmath157 . * proof . *
if @xmath154}$ ] and @xmath155}$ ] then @xmath131 contains the pattern @xmath158 ; we apply to @xmath131 the following operation @xmath159 : * consider the straight line @xmath136 from the end of the pattern @xmath56 and let @xmath160 be the leftmost point where @xmath136 intersects @xmath131 on the right of @xmath56 such that @xmath160 is followed by at least two rise steps ; * let @xmath139 , @xmath140 , the subsequence on the left of @xmath160 , preceded by at least a rise step ; * swap the subsequence @xmath56 and @xmath142 .
when @xmath143 , that is @xmath142 is the empty word , we simply insert @xmath56 into @xmath160 .
@xmath153 figure [ fig : tree ] shows the initial steps of the generating algorithm of the paths corresponding to words in @xmath9}$ ] , @xmath161 .
let us remark that , following the above constructions , given a path @xmath4 , the number of generated paths depends only on the ordinate of endpoint of @xmath4 .
so , the complete generating algorithm can be briefly described by the succession rule ( [ one ] ) ( for more details on succession rules see @xcite ) @xmath162 where each number corresponds to the ordinate of the endpoint of a path .
the zero in the first line in ( [ one ] ) is associated with the empty path . the second line in ( [ one ] )
is associated with operations @xmath163 and @xmath164 , the third line is associated with operations @xmath165 and @xmath166 , and the last line describes the construction when the endpoint has ordinate @xmath167 , underground path included .
in this section we prove that the construction described in section 3 allows to generate the class @xmath9}$ ] exhaustively for any fixed forbidden pattern @xmath17 , @xmath19 , in the sense that all the words in @xmath9}$ ] with @xmath31 1 s , @xmath168 , can be generated .
[ teo : uno ] given a fixed forbidden pattern @xmath17 , @xmath169 , the construction described in section [ sec : algo ] generates all the paths with @xmath31 , @xmath168 , rise steps representing the binary words in @xmath9}$ ] with @xmath31 1 s .
let @xmath11}$ ] , then @xmath170 that is , @xmath4 is made of @xmath171 sub - paths such that : * @xmath172 is the empty path @xmath40 , * @xmath173 , @xmath174 is a path in @xmath9}$ ] beginning from and ending on the @xmath41-axis , * @xmath175 is a path in @xmath9}$ ] beginning from the @xmath41-axis with endpoint at ordinate @xmath176 . the proof of theorem [ teo : uno ] is obtained by induction on the number of sub - paths . * proof . *
the empty path @xmath40 is generated as the starting point of the algorithm .
let us assume that all the possible sub - paths @xmath177 of @xmath4 are generated .
we prove that the algorithm generates the sub - path @xmath178 for any path @xmath179}$ ] .
let @xmath173 be a path that does not end with the pattern @xmath56 . in this case
the path @xmath180 may be either positive or negative . if @xmath180 is a positive path we have to prove that all the positive paths are generated and this will be demonstrated in section [ sec : positive ] . in the case of a negative path , denoting @xmath181 the largest value of its absolute ordinate , let us remark that : * the negative paths with @xmath182 are @xmath183 , @xmath184 , and they are generated by iterating the construction ( [ alfa ] ) ; * negative paths with @xmath185 are @xmath186 , @xmath187 , and they are generated by means of ( [ alfa ] ) when @xmath188 , or by means of ( [ alfa1 ] ) when @xmath189 ; * let @xmath190 be a negative path with @xmath191 ; if @xmath192}$ ] then @xmath190 is the underground path generated by a positive path with endpoint at ordinate @xmath55 , ( see section [ sec : under ] ) , otherwise @xmath193 for a positive path @xmath83 in @xmath9}$ ] with endpoint at ordinate @xmath194 ( see proposition [ prop3 ] ) .
note that when @xmath195 and @xmath173 ends with the pattern @xmath88 , the only possible negative path @xmath180 with @xmath185 is @xmath196 and it is generated by applying the construction ( [ alfa1 ] ) to the path @xmath197 . when the suffix of @xmath173 is the pattern @xmath56 , @xmath180 must be a negative path and the path @xmath198 is the underground path obtained by means of the construction described in case 1 ) in section [ sec : negativesuffix ] ( see figure [ fig : casoa ] ) . in the same way , when the sub - path @xmath199 is of type @xmath200 or @xmath201 where @xmath113 is
the empty path @xmath40 or is a strongly negative path , then it is generated by the constructions described in case 2 ) in section [ sec : negativesuffix ] ( see figures [ fig : casob1 ] , [ fig : casob2i ] and [ fig : casob2ii ] ) .
then , if we show that all the possible positive paths are generated , then we can claim that theorem [ teo : uno ] is proved .
moreover , we observe that for each path @xmath4 in @xmath9}$ ] with @xmath31 rise steps there exists one and only one path @xmath125 in @xmath9}$ ] with @xmath202 rise steps , @xmath51 , such that @xmath4 is obtained from @xmath125 by means of the construction described in section [ sec : algo ] .
this assertion is a direct consequence of the construction , since the actions described are univocally determined .
@xmath153 in this section we prove that all the positive paths with @xmath31 rise steps are generated by means of the construction described in section [ sec : algo ] . in the sequel of this section
we analyze only _ positive _ paths .
the proof is obtained by induction on @xmath31 .
there are only two paths with @xmath203 rise step , that is @xmath41 and @xmath45 , and they are generated by means of construction ( [ alfa ] ) applied to the empty path @xmath40 .
let us assume that all the paths with @xmath204 rise steps are generated ; we will prove that all the paths with @xmath31 rise steps are generated .
note that , following the construction given in @xcite , a path with @xmath31 rise steps can be obtained from a dyck path @xmath4 with @xmath205 rise step by inserting one rise step in each point at ordinate @xmath206 of its last descent followed by @xmath207 fall steps , @xmath208 .
we will prove that all the paths obtained so are also generated following the constructions given in the above sections .
let us denote by @xmath209 a paths ending with @xmath207 fall steps .
let @xmath69 be the number of fall steps in the last descent of @xmath4 .
first of all , we note that for any value of @xmath69 the paths obtained by inserting a rise step in the point at ordinate 0 are generated by means of the constructions ( [ alfa ] ) or ( [ alfa1 ] ) applied to the path @xmath4 . here
we give the proof for the case with @xmath210 , distinguishing three cases : @xmath211 , @xmath212 and @xmath213 .
the analogous and simple cases @xmath214 and @xmath215 are left to the reader .
* @xmath216 .
let @xmath217 .
the insertion of a rise step in the point at ordinate 1 gives three paths : * * @xmath218 , which is generated by means of ( [ beta ] ) applied to the prefix @xmath219 of @xmath4 , * * @xmath220 and @xmath221 , which are the paths with endpoints at ordinate 1 and 0 , respectively .
if @xmath222 , then @xmath223 and @xmath224 are generated by the construction ( [ gamma ] ) , where @xmath225 , applied to the path @xmath226 with @xmath227 , otherwise , if @xmath195 , they are generated by means of ( [ gamma1 ] ) with @xmath225 and @xmath85 applied to the path @xmath228 . *
the insertion of a rise step in the point at ordinate @xmath206 gives @xmath230 paths @xmath209 ,
@xmath231 . the paths @xmath209 with @xmath232 are all the positive paths generated by means of ( [ gamma ] ) with @xmath233 and @xmath85 applied to the prefix of @xmath4 of length @xmath234 .
the path @xmath235 is the path with endpoint at ordinate @xmath206 .
when @xmath236 , @xmath223 is generated by means of construction ( [ gamma ] ) with @xmath225 and @xmath227 applied to the path @xmath226 , while , when @xmath195 , it is generated by means of ( [ gamma1 ] ) , with @xmath225 and @xmath85 , applied to the path @xmath228 .
* @xmath213 .
the insertion of a rise step in the point at ordinate @xmath237 generates @xmath238 paths @xmath209 ,
@xmath239 . the paths @xmath209 with @xmath232 are all the positive paths generated by means of ( [ gamma ] ) with @xmath225 and @xmath85 applied to the prefix of @xmath4 of length @xmath240 .
as far as the generation of @xmath223 is concerned , we have to distinguish three cases : 1 .
if @xmath241 , @xmath242 , then @xmath243 . if @xmath244 , then @xmath223 is the path with endpoint at ordinate @xmath66 generated by means of ( [ gamma ] ) with @xmath245 and @xmath246 , applied to the path @xmath247 .
if @xmath248 , then @xmath223 is the path with endpoint at ordinate @xmath66 generated by means of ( [ gamma1 ] ) with @xmath245 and @xmath85 applied to the path @xmath249 .
note that , when @xmath250 , the endpoint of the prefix @xmath247 has ordinate 1 , and the path @xmath223 is obtained applying the construction ( [ beta ] ) ( or ( [ beta1 ] ) ) instead of ( [ gamma ] ) ( or ( [ gamma1 ] ) ) .
if @xmath251 , @xmath252 and @xmath253 , then @xmath254 . if @xmath244 , then @xmath223 is the path with endpoint at ordinate @xmath255 generated by means of ( [ gamma ] ) with @xmath256 and @xmath257 applied to the path @xmath258 .
if @xmath248 , then @xmath223 is generated by means of ( [ gamma1 ] ) with @xmath256 and @xmath85 applied to the path @xmath259 .
if @xmath260 , @xmath261 and @xmath252 , then @xmath262 . if @xmath244 , then @xmath223 is the path with endpoint at ordinate @xmath66 generated by means of ( [ gamma ] ) , with @xmath245 and @xmath257 applied to the path @xmath263 . if @xmath248 , then @xmath223 is generated by means of ( [ gamma1 ] ) , with @xmath245 and @xmath264 applied to the path @xmath265 .
in this paper we propose an algorithm for the construction of particular binary words , according to the number of 1 s , excluding a fixed pattern @xmath266 , @xmath19 .
moreover , it would be interesting to study words avoiding patterns which have a different shape , that is not only patterns consisting of a sequence of rise and fall steps
. this could be the first step in the study of a possible universal generating algorithm for pattern avoiding words .
another interesting field of study is to determine a sort of invariant class of avoiding patterns that is the paths @xmath267 such that @xmath268}|=|f^{[{\mathfrak{p } } _
2]}|= \dots = |f^{[{\mathfrak{p } } _ l]}|$ ] with consequent bijective problems .
one could also consider a forbidden pattern on an arbitrary alphabet and investigate words avoiding that pattern , or study words avoiding more than one pattern and the related combinatorial objects , considering various parameters .
s. bilotta , d. merlini , e. pergola , r. pinzani , binary words avoiding a pattern and marked succession rule , _ lattice path combinatorics and applicatons , siena , july 4 - 7 , 2010 _ ( available on line arxiv:1103.5689 ) , ( 2010 ) . | in this paper we propose an algorithm to generate binary words with no more 0 s than 1 s having a fixed number of 1 s and avoiding the pattern @xmath0 for any fixed @xmath1 .
we will prove that this generation is exhaustive , that is , all such binary words are generated .
* keywords : * binary words , pattern avoiding , exhaustive generation . |
at least 10% of all neutron stars are born as magnetars , with magnetic fields @xmath5 g. their activity is powered by the decay of the ultrastrong field and lasts about @xmath6 years .
they are observed at this active stage as either soft gamma repeaters ( sgrs ) or anomalous x - ray pulsars ( axps ) @xcite . besides the sporadic x - ray outbursts , a second , persistent
, form of activity has been discovered by studying the emission spectra of magnetars . until recently ,
the spectrum was known to have a thermal component with temperature @xmath7 kev , interpreted as blackbody emission from the star s surface .
the soft x - ray spectrum also showed a tail at @xmath8 kev with photon index @xmath9 .
this deviation from pure surface emission already suggested that energy is partially released above the star s surface .
recently , observations by rxte and integral have revealed even more intriguing feature : magnetars are bright , persistent sources of 100 kev x - rays @xcite .
this high - energy emission forms a separate component of the magnetar spectrum .
it becomes dominant above 10 kev , has a very hard spectrum , @xmath10 , and peaks above 100 kev where the spectrum is unknown .
its luminosity , @xmath11 erg s@xmath12 , even exceeds the thermal luminosity from the star s surface .
the observed hard x - rays can be emitted only in the exterior of the star and demonstrate the presence of an active plasma corona .
this corona can form as a result of starquakes that shear the magnetar crust and its external magnetic field .
the persistence of magnetospheric twists can explain several observed properties of magnetars .
thompson , lyutikov , & kulkarni @xcite investigated the observational consequences using a static , force - free model , idealizing the magnetosphere as a globally twisted dipole .
they showed that a twist affects the spindown rate of the neutron star : it causes the magnetosphere to flare out slightly from a pure dipole configuration , thereby increasing the braking torque acting on the star .
although the calculations of force - free configurations are independent of the plasma behavior in the magnetosphere , they rely on the presence of plasma that can conduct the required current @xmath13 .
this requires a minimum particle density @xmath14 .
thompson et al .
@xcite showed that even this minimum density can modify the stellar radiation by multiple resonant scattering .
the problem of plasma dynamics in the closed twisted magnetosphere has been formulated in @xcite , and a simple solution has been found to this problem .
it allows one to understand the observed energy output of the magnetar corona and the evolution of magnetic twists .
it is worth mentioning that plasma behavior around neutron stars has been studied for decades in the context of radio pulsars , and that problem remains unsolved .
the principle difference with radio pulsars is that their activity is caused by _
rotation _ , and dissipation occurs on _ open magnetic lines _ that connect the star to its light cylinder .
electron - positron pairs are then created on open field lines @xcite . by contrast , the formation of a corona around a magnetar does not depend on its rotation . all observed magnetars are slow rotators ( @xmath15 hz ) , and their rotational energy is unable to feed the observed coronal emission .
practically the entire plasma corona is immersed in the _ closed _ magnetosphere and its heating must be caused by some form of dissipation on the closed field lines .
it is this closure that facilitates the corona problem by providing both boundary conditions at the two footpoints of a field line .
the basic questions that one would like to answer are as follows .
how is the magnetosphere populated with plasma ?
the neutron star surface has a temperature @xmath16 kev and the scale - height of its atmospheric layer ( if it exists ) is only a few cm .
how is the plasma supplied above this layer and what type of particles populate the corona
? how is the corona heated , and what are the typical energies of the particles ?
if the corona conducts a current associated with a magnetic twist @xmath17 , how rapid is the dissipation of this current , i.e. what is the lifetime of the twist ? does its decay imply the disappearance of the corona ?
an outline of the proposed model is as follows .
the key agent in corona formation is an electric field @xmath18 parallel to the magnetic field .
@xmath18 is generated by the self - induction of the gradually decaying current and in essence measures the rate of the decay .
it determines the heating rate of the corona via joule dissipation . if @xmath19 then the corona is not heated and , being in contact with the cool stellar surface , it will have to condense to a thin surface layer with @xmath16 kev .
the current - carrying particles can not flow upward against gravity unless a force @xmath20 drives it . on the other hand ,
when @xmath18 exceeds a critical value , @xmath21 avalanches are triggered in the magnetosphere , and the created pairs screen the electric field .
this leads to a `` bottleneck '' for the decay of a magnetic twist , which implies a slow decay .
maintenance of the corona and the slow decay of the magnetic twist are intimately related because both are governed by @xmath18 . in order to find @xmath18
, one can use gauss law @xmath22 where @xmath23 is the net charge density of the coronal plasma .
this constraint implies that @xmath24 and @xmath23 must be found self - consistently .
the problem turns out to be similar to the classical double - layer problem of plasma physics with a new essential ingredient : @xmath21 creation .
a direct numerical experiment can be designed that simulates the creation and behavior of the plasma in the magnetosphere .
the experiment shows how the plasma and electric field self - organize to maintain the time - average magnetospheric current @xmath25 demanded by the magnetic field , @xmath26 .
the electric current admits no steady state on short timescales and keeps fluctuating , producing @xmath21 avalanches .
this state may be described as a self - organized criticality .
pair creation is found to provide a robust mechanism for limiting the voltage along the magnetic lines to @xmath27 gev and regulate the observed luminosity of the corona .
a tightly wound - up magnetic field is assumed to exist inside magnetars @xcite .
the internal toroidal field can be much stronger than the external large - scale dipole component .
the essence of magnetar activity is the transfer of magnetic helicity from the interior of the star to its exterior , where it dissipates .
this involves rotational motions of the crust , which inevitably twist the external magnetosphere anchored to the stellar surface .
the magnetosphere is probably twisted in a catastrophic way , when the internal magnetic field breaks the crust and rotates it .
such starquakes are associated with observed x - ray bursts @xcite .
the most interesting effect of a starquake for us here is the partial release of the winding of the internal magnetic field into the exterior , i.e. , the injection of magnetic helicity into the magnetosphere .
since the magnetic field is energetically dominant in the magnetosphere , it must relax there to a force - free configuration with @xmath28 .
electric current @xmath29 initiated by a starquake flows along the magnetic field lines to the exterior of the star , reaches the top of the field line , and comes back to the star at the other footpoint , then enters the atmosphere and the crust , following the magnetic field lines ( @xmath30 ) .
sufficiently deep in the crust the current flows across the field lines and changes direction so that it can flow back to the twisted footpoint through the magnetosphere ( or directly through the star , depending on the geometry of the twist ) .
the current can flow across the field lines in the deep crust because it is able to sustain a significant ampre force @xmath31 this force is balanced by elastic forces of the deformed crust .
the emerging currents are easily maintained during the x - ray outburst accompanying a starquake .
a dense , thermalized plasma is then present in the magnetosphere , which easily conducts the current .
plasma remains suspended for some time after the starquake because of the transient thermal afterglow @xcite .
eventually the afterglow extinguishes and the radiative flux becomes unable to support the plasma outside the star .
the decreasing density then threats the capability of the magnetosphere to conduct the current of the magnetic twist . a minimal `` corotation '' charge density @xmath32 is always maintained @xcite , but it is far smaller than needed to supply the current @xmath33 .
can the lack of charges stop the flow of electric current ?
a simple estimate shows that the current can not stop quickly because of its self - induction . a slow decay of the current generates a sufficient self - induction voltage that helps the magnetosphere to re - generate the plasma that carries the current , so that the twisted force - free configuration persists with a qausi - steady @xmath34 .
the stored energy of non - potential ( toroidal ) magnetic field associated with the ejected current is subject to gradual dissipation . in our model
, this dissipation feeds the observed activity of the corona .
the stored energy is concentrated near the star and carried by closed magnetic lines with a maximum extension radius @xmath35 .
thus , most of the twist energy will be released if these lines untwist , and we focus here on the near magnetosphere @xmath35 .
consider a magnetic flux tube with cross section @xmath36 and length @xmath37 which carries a current @xmath38 .
the stored magnetic energy of the current per unit length of the tube is ~ s. the decay of this energy is associated with an electric field parallel to the magnetic lines @xmath39 : this field can accelerate particles and convert the magnetic energy into plasma energy .
conservation of energy can be expressed as ( ) = - -(c ) , as follows from maxwell s equations with @xmath40 .
the first term on the right - hand side is the joule dissipation caused by @xmath39 .
the second term the divergence of the poynting flux vanishes for the induced field @xmath41 .
is a few cm.,scaledwidth=48.0% ] the decay of the twist is related to the voltage between the footpoints of a magnetic line , [ eq : phi_e ] _
e=_a^c e. here a and c are the anode and cathode footpoints and @xmath42 is the line element ; the integral is taken along the magnetic line outside the star ( fig . 1 ) .
the product @xmath43 approximately represents the dissipation rate @xmath44 in the tube .
( the two quantities are not exactly equal in a time - dependent circuit . )
the instantaneous dissipation rate is given by = _
a^c e(l ) , where @xmath45 is the instantaneous current at position @xmath46 in the tube .
the current is fixed at footpoints a and c , @xmath47 ; it may , however , fluctuate between the footpoints .
the voltage @xmath48 is entirely maintained by the self - induction that accompanies the gradual untwisting of the magnetic field .
thus , @xmath48 reflects the gradual decrease of the magnetic helicity in the tube , = t_a^c * a * v = -_a^c * eb * s(l ) l. ( here @xmath49 is the volume element . ) it also determines the effective resistivity of the tube : @xmath50 .
the self - induction voltage passes the released magnetic energy to charged particles , and a higher rate of untwisting implies a higher energy @xmath51 gained per particle .
a huge magnetic energy is stored in the twisted tube , and a quick untwisting would lead to extremely high lorentz factors of the accelerated particles .
there is , however , a bottleneck that prevents a fast decay of the twist : the tube responds to high voltages through the copious production of @xmath21 pairs .
runaway pair creation is ignited when electrons are accelerated to a certain lorentz factor @xmath52 ( see below ) .
the created @xmath21 plasma does not tolerate large @xmath18 the plasma is immediately polarized , screening the electric field .
this provides a negative feedback that limits the magnitude of @xmath48 and buffers the decay of the twist .
a minimum @xmath48 is needed for the formation of a corona .
two mechanisms can supply plasma : ( 1 ) ions and electrons can be lifted into the magnetosphere from the stellar surface .
this requires a minimum voltage , e_e ~gm_i r_ns ~200 , corresponding to @xmath18 that is strong enough to lift ions ( of mass @xmath53 ) from the anode footpoint and electrons from the cathode footpoint .
( 2 ) pairs can be created in the magnetosphere if e_e ~_mc^2 = 0.5 _ , which can accelerate particles to the lorentz factor @xmath54 sufficient to ignite @xmath21 creation .
if @xmath48 is too low and the plasma is not supplied , a flux tube is guaranteed to generate a stronger electric field . the current becomes slightly charge - starved : that is , the net density of free charges becomes smaller than @xmath55 .
the ultrastrong magnetic field , whose twist carries an enormous energy compared with the energy of the plasma , does not change and responds to the decrease of @xmath29 by generating an electric field : a small reduction of the conduction current @xmath29 induces a displacement current @xmath56 .
the longitudinal electric field @xmath18 then quickly grows until it can pull particles away from the stellar surface and ignite pair creation .
the limiting cases @xmath57 ( no decay of the twist ) and @xmath58 ( fast decay ) both imply a contradiction . the electric field and
the plasma content of the corona must regulate each other toward a self - consistent state , and the gradual decay of the twist proceeds through a delicate balance : @xmath18 must be strong enough to supply plasma and maintain the current in the corona ; however , if plasma is oversupplied @xmath18 will be reduced by screening . a cyclic behavior is possible , in which plasma is periodically oversupplied and @xmath18 is screened .
our numerical experiment shows that such a cyclic behavior indeed takes place .
three facts facilitate the simulation of the coronal electric circuit : \1 .
the ultrastrong magnetic field makes the particle dynamics 1-d .
the magnetic field lines are not perturbed significantly by the plasma inertia , and they can be thought of as fixed `` rails '' along which the particles move .
the particle motion is confined to the lowest landau level and is strictly parallel to the field ( the lifetime of a particle in an excited landau state is tiny ) .
the particle motion is collisionless in the magnetosphere .
it is governed by two forces only : the component of gravity projected onto the magnetic field and a collective electric field @xmath18 which is determined by the charge density distribution and must be found self - consistently .
the star possesses a dense and thin atmospheric layer . near the base of the atmosphere ,
the required current @xmath59 is easily conducted , with almost no electric field .
therefore the circuit has simple boundary conditions : @xmath19 and fixed current .
the atmosphere is much thicker than the skin depth of the plasma and screens the magnetospheric electric field from the star .
our goal is to understand the plasma behavior above the screening layer , where the atmospheric density is exponentially reduced and an electric field @xmath18 must develop .
the induced electric field @xmath24 and conduction current @xmath29 satisfy the maxwell equation , [ eq : maxw _ ] = + . here
@xmath29 is parallel to the direction of the magnetic field , and the force - free condition requires @xmath60 .
therefore , in the fixed magnetic configuration where @xmath61 does not vary with time , @xmath62 is parallel to @xmath63 , i.e. only @xmath18 is created by the self - iduction effect .
we consider a simple model problem where no pre - existing perpendicular field @xmath64 exists ; more precisely , we require @xmath65 .
then gauss law reads [ eq : gauss1 ] 4=_= , where @xmath46 is length measured along the magnetic tube .
then the problem becomes strictly 1-d since @xmath64 has no relation to charge density and falls out from the problem .
note that the approximation of 1d circuit , @xmath66 , where @xmath67 , excludes the excitation of transverse waves in the magnetosphere , which in reality can exist .
these waves are described by the coupled fluctuations @xmath68 and @xmath64 on scales much smaller than the circuit size @xmath69 .
( wavelengths @xmath70 can be excited by plasma oscillations . )
these fluctuations may be expected to have a small effect on the circuit solution unless @xmath68 becomes comparable to @xmath71 . with @xmath66 ,
the particle motions on different magnetic lines are decoupled .
indeed , the particle dynamics is controlled by electric field @xmath72 which is related to the instantaneous charge distribution by gauss law ( [ eq : gauss1 ] ) .
thus , @xmath73 on a given magnetic line is fully determined by charge density @xmath74 _ on the same line_. the plasma and electric field evolve along the line as if the world were 1-d .
if the conduction current @xmath75 is smaller than @xmath71 , then @xmath76 and an electric field appears that tends to increase the current . alternatively , if @xmath77 then an electric field of the opposite sign develops which tends to reduce the current .
thus , @xmath75 is always regulated toward @xmath78 .
this is the standard self - induction effect .
the timescale of the regulation , @xmath79 , is very short , of the order of the langmuir oscillation timescale @xmath80 s. local deviations from charge neutrality tend to be erased on the same plasma timescale .
the net charge imbalance of the corona must be extremely small a tiny imbalance would create an electric field that easily pulls out the missing charges from the surface . when the corotation charge density is neglected , the net charge of the corona vanishes . adopting a characteristic velocity @xmath81 of the coronal particles , the debye length of the plasma @xmath82
can be taken equal to the plasma skin depth @xmath83 .
an important physical parameter of the corona is the ratio of the debye length to the circuit size @xmath37 , [ eq : zeta ] = = 1 .
the force - free condition @xmath84 together with @xmath85 requires that @xmath86 along the magnetic line .
we can scale out this variation simply dividing all local quantities ( charge density , current density , and electric field ) by @xmath87 .
this reduces the problem to an equivalent problem where @xmath88 .
furthermore , only forces along magnetic lines control the plasma dynamics , and the curvature of magnetic lines falls out from the problem
. therefore , we can set up the experiment so that plasma particles move along a straight line connecting anode and cathode ( fig . 2 ) .
we designated this line as the @xmath89-axis , so that @xmath90 .
traps the particles near the boundaries .
the constant current at the boundaries , however , implies that voltage is immediately generated should the flow of charge stop in the tube .
the experiment aims to find the self - induction voltage that keeps the current flowing .
, scaledwidth=48.0% ] this one - dimensional system may be simulated numerically .
consider @xmath91 particles moving along the @xmath89-axis ( @xmath92 in our simulations ) .
the electric field acting on a particle at a given @xmath89 is given by [ eq : gauss ] e(z)=4_0^zz , where @xmath93 are the boundary conditions .
given the instantaneous positions @xmath94 of all charges @xmath95 we immediately find @xmath96 . to find a quasi - steady state of such a system we let it relax by following the particle motion in the self - consistent electric field .
consider first a circuit where @xmath21 production is ( artificially ) forbidden , so that only the cold hydrostatic atmosphere can supply particles to the corona .
we find that such a circuit quickly relaxes to a state where it acts as an ultra - relativistic linear accelerator ( fig . 3 ) .
in this state , oscillations of electric field ( on the plasma timescale @xmath97 ) are confined to the thin atmospheric layers .
a static accelerating electric field is created above the layers where the atmosphere density is exponentially reduced .
the asymmetry of the solution is caused by the difference between the electron and ion masses .
upper panel shows the normalized charge density and lower panel shows the electric field in units of @xmath98 defined by @xmath99 .
solid and dotted curves correspond to two different moments of time , demonstrating that a quasi - steady state has been reached . in this simulation , @xmath100 and @xmath101 .
the final state is the relativistic double layer described analytically in @xcite .
, scaledwidth=52.0% ] this configuration is a relativistic double layer @xcite .
it is well described by carlqvist s solution , which has no gravity in the circuit and assumes zero temperature at the boundaries , so that particles are injected with zero velocity . according to this solution ,
the potential drop between anode and cathode , in the limit @xmath102 , is [ eq : phic ] e_e = m_ec , _
p^2 , where @xmath103 is the ion charge number ( in our experiment @xmath104 was assumed ) .
the established voltage is much larger than is needed to overcome the gravitational barrier @xmath105 .
it does not even depend on @xmath105 as long as @xmath105 is large enough to prohibit the thermally - fed regime .
gravity causes the transition to the linear - accelerator state , but the state itself does not depend on @xmath105 .
thus , the first result of numerical experiment is that an ion - electron circuit ( without @xmath21 creation ) in a gravitational field relaxes to the double layer of macroscopic size @xmath69 and huge voltage @xmath48 .
the system does not find any state with a lower @xmath48 , even though it is allowed to be time - dependent and strong fluctuations persist in the atmospheric layers ( reversing the sign of @xmath106 ) .
if the linear accelerator were maintained in the twisted magnetosphere , the twist would be immediately killed off : the huge voltage implies a large untwisting rate ( 2 ) .
the electron lorentz factor developed in the linear accelerator is ( taking the real @xmath107 and @xmath108),[eq : gamma ] _
e= ~610 ^ 9 .
however , new processes will become important before the particles could acquire such high energies : production of @xmath21 pairs will take place .
therefore , the linear - accelerator solution can not describe a real magnetosphere .
we conclude that pair creation is a key ingredient of the circuit that will regulate the voltage to a smaller value .
in the magnetospheres of canonical radio pulsars with @xmath109 g , @xmath21 pairs are created when seed electrons are accelerated to large lorentz factors @xmath110 .
such electrons emit curvature gamma rays that can convert to @xmath21 off the magnetic field . in stronger fields ,
another channel of @xmath21 creation appears .
it is also two - step : an accelerated particle resonantly upscatters a thermal x - ray photon , which subsequently converts to a pair @xcite .
the resonant scattering requires the electron to have a lorentz factor , [ eq : gres ] = b / b_qed1-_kb 10 ^ 3b_15 , where @xmath111 and @xmath112 are the energy and pitch angle of the target photon .
the distance over which an accelerated electron creates a pair is approximately equal to the free path to scattering @xmath113 .
it depends only on @xmath87 and the spectrum of target photons ; @xmath114 for the relevant parameter @xmath115 kev @xcite .
when the voltage in the circuit becomes high enough to accelerate electrons to @xmath116 , an @xmath21 breakdown develops in the magnetic flux tube .
an illustrative toy model of breakdown is shown in a spacetime diagram in figure 4 .
each pair - creation event gives two new particles of opposite charge , which initially move in the same direction .
one of them is accelerated by the electric field and lost after a time @xmath117 , while the other is decelerated and can reverse direction before reaching the boundary .
this reversal of particles in the tube and repeated pair creation allows the @xmath21 plasma to be continually replenished . in the super - critical regime ( left panel in fig .
4 ) more than one reversing particle is created per passage time @xmath118 and an avalanche develops exponentially on a timescale @xmath119 . in the near - critical regime shown in the right panel ,
just one reversing particle is created per passage time @xmath118 .
this critical state is unstable : sooner or later the avalanche will be extinguished .
breakdown and formation of avalanches .
the seed electron is placed at cathode and is accelerated to @xmath116 after passing the distance @xmath120 .
the figure shows the worldlines of created particles .
( a ) @xmath121 ( supercritical case ) .
( b ) @xmath122 ( critical case ) .
, scaledwidth=52.0% ] this toy model shows an essential property of the @xmath21 breakdown : it is a critical stochastic phenomenon . above a critical voltage pair creation proceeds in a runway manner , and the current and
the dissipation rate would run away if the voltage were fixed . below the critical voltage pair creation does not ignite .
the criticality parameter is @xmath123 .
the tube with enforced current at the boundaries must self - organize to create pairs in the near - critical regime @xmath124 and maintain the current .
the discharging tube is similar to other phenomena that show self - organized criticality , e.g. , a pile of sand on a table @xcite .
if sand is steadily added , a quasi - steady state is established with a characteristic mean slope of the pile .
the sand is lost ( falls from the table ) intermittently , through avalanches a sort of `` sand discharge . '' in our case , charges of the opposite signs are added steadily instead of sand ( fixed @xmath75 at the boundaries ) , and voltage @xmath125 plays the role of the mean slope of a pile .
the behavior of the discharging system is expected to be time - dependent , with stochastic avalanches .
we implemented the process of pair creation in our numerical experiment .
regardless the details of this process ( values of @xmath116 and @xmath113 ) the circuit relaxed to the state of self - organized criticality after a few @xmath126 .
this state is time - dependent on short timescales , but it has a well - defined steady voltage when averaged over a few @xmath127 ( fig . 5 ) ,
we found that this relation applies even to circuits with @xmath128 , where lifting of the ions is energetically preferable to pair creation .
we conclude that _ the voltage along the magnetic tube is self - regulated to a value just enough to maintain pair production and feed the current with @xmath21 pairs_. in all cases , the pair creation rate @xmath129 is maintained .
the solution for the plasma dynamics in the corona is strongly non - linear and time - dependent .
it is essentially global in the sense that the plasma behavior near one footpoint of a magnetospheric field line is coupled to the behavior near the other footpoint .
remarkably , this complicated global behavior may be described as essentially one - dimensional electric circuit that is subject to simple boundary conditions @xmath130 on the surface of the star and can be studied using a direct numerical experiment . , @xmath131 , @xmath132 .
after a few dynamical times the voltage @xmath48 stops growing and the circuit enters an oscillatory regime . during each oscillation , an increased voltage triggers @xmath21 discharge , then @xmath133 drops , and @xmath48 begins to grow again . the quasi - steady oscillating state ( self - organized criticality ) is established .
the horizontal lines show the voltage @xmath134 and the minimum pair production rate @xmath135 that can feed the required current @xmath71 .
, scaledwidth=52.0% ] the current is carried largely by @xmath21 everywhere in the tube .
the ion fraction in the current depends on the ratio @xmath136 ( fig .
6 ) . if this ratio is small , pairs are easily produced with a small electric field , too small to lift the ions , and the ion current is suppressed . in the opposite case ,
ions carry about 1/2 of the current . by coincidence
, @xmath137 is typical for magnetars , which implies that ions carry @xmath138% of the current . , and in the right panel in units of @xmath139 .
different colors and symbols show circuits with @xmath140 and various values of @xmath116 ranging from @xmath141 to @xmath142 ( see @xcite ) .
, scaledwidth=49.0% ]
the basic finding of this work is that an @xmath21 corona must be maintained around a magnetar .
it exists in a state of self - organized criticality , near the threshold for @xmath21 breakdown .
the stochastic @xmath21 discharge continually replenishes the coronal plasma , which is lost on the dynamic time @xmath143 s. the established voltage along a twisted magnetic tube is marginally sufficient to accelerate an electron ( or positron ) to the energy where collisions with ambient x - ray photons spawn new pairs .
this physical condition determines the rate of energy dissipation in the corona .
the generation of this voltage is precisely the self - induction effect of the gradually decaying magnetic twist , and the energy release in the corona is fed by the magnetic energy of the twisted field .
the rate of energy dissipation in the twisted magnetosphere is given by @xmath144 where @xmath145 is the net current through the corona .
the current may be estimated as @xmath146 , where @xmath147 is the size of a twisted region on the stellar surface and @xmath148 characterizes the strength of the twist .
the calculated @xmath149 gev implies [ eq : ldiss ] l_diss = i_e~10 ^ 37b_15 ( ) ^2 ( ) . observed luminosity @xmath150 erg s@xmath12 is consistent with a partially twisted magnetosphere , @xmath151 , or a global moderate twist with @xmath152 .
once created , a magnetospheric twist has a relatively long but finite lifetime .
the energy stored in it is @xmath153 .
this energy dissipates in a time @xmath154 note that a stronger twist ( brighter corona ) lives longer .
if the time between large - scale starquakes is longer than @xmath155 , the magnetar should be seen to enter a quiescent state .
such behavior has been observed in axp j1810 - 197 : an outburst was followed by the gradual decay on a year timescale @xcite .
the voltage @xmath48 implies a certain effective resistivity of the corona , @xmath50 .
this resistivity leads to spreading of the electric current _ across _ the magnetic lines , which is described by the induction equation @xmath156 .
the timescale of twist spreading to the light cylinder is comparable to @xmath155 .
therefore , the impact of a starquake on spindown may appear with a delay of @xmath157 years .
even a small ion current implies a large mass transfer through the magnetosphere over the @xmath158-yr active lifetime of a magnetar .
it requires a significant excavation of the crust at the anode footpoints of twisted magnetic tubes .
the bombarding relativistic electrons from the corona cause this excavation .
they spall heavy ions in the uppermost crust , re - generate the light - element atmosphere on the surface , and regulate its column density to a value @xmath159 @xcite . the radiative output from the corona likely peaks in its inner region because the coronal current is concentrated on closed field lines with a maximum extension @xmath160 ( 2 ) .
emission from the inner corona must be suppressed above @xmath3 mev , regardless the mechanism of emission , because photons with energy @xmath161 mev can not escape the ultra - strong magnetic field .
the observed high - energy radiation extends above 100 kev @xcite .
there is an indication for a cutoff between 200 kev and 1 mev from comptel upper limits for axp 4u 0142@xmath16261
. the possible mechanisms of emission are strongly constrained .
the corona has a low density @xmath163 @xmath164 and particle collisions are rare , so two - body radiative processes are negligible .
the upscattering of kev photons streaming from the surface does not explain the observed 100-kev emission @xcite .
a possible source of the 100 kev x - rays is the transition layer between the corona and the thermal photosphere @xcite .
it is heated by the coronal beam and cooled by heat conduction toward the surface , and its temperature is regulated to @xmath165 kev , where @xmath166 parameterizes the suppression of thermal conductivity by plasma turbulence .
the emitted x - ray spectrum may be approximated as single - temperature optically thin bremsstrahlung @xcite .
its photon index below the exponential cutoff is close to @xmath167 , in agreement with spectra observed with integral and rxte .
the observed pulsed fraction increases toward the high - energy end of the spectrum and approaches 100% at 100 kev ( kuiper et al .
if the hard x - rays are produced by one or two twisted spots on the star surface , the large pulsed fraction implies that the spots almost disappear during some phase of rotation
. then they should not be too far from each other .
they can not be , e.g. , antipodal because a large fraction of the star surface is visible to observer due to the gravitational bending of light : @xmath168 @xcite .
the transition layer at a footprint of a coronal flux tube may reach high enough temperature for thermal @xmath21 production @xcite .
the created pairs then elevate above the surface and can conduct the magnetospheric current .
this opens a possibility for the corona to be fed by surface pair creation instead of the @xmath21 discharge .
the observed infrared ( @xmath169-band ) and optical luminosities of magnetars are @xmath170 erg / s @xcite , which is far above the rayleigh - jeans tail of the surface blackbody radiation .
the inferred brightness temperature is @xmath171 k if the radiation is emitted from the surface of the star .
four principle emission mechanisms may then be considered : coherent plasma emission , synchrotron emission by electrons , cyclotron emission by ions , and curvature radiation by @xmath21 bunches in the corona .
the last two are most plausible ( see @xcite for detailed discussion ) .
in particular , curvature radiation in the inner magnetosphere may extend into the optical - ir range .
charges moving on field lines with a curvature radius @xmath172 will emit radiation with frequency @xmath173 , which is in the k - band ( @xmath174 hz ) or optical when @xmath175 gev .
the radio luminosity of a normal pulsar can approach @xmath176 of its spindown power , and it is not implausible that the observed fraction @xmath2 of the power dissipated in a magnetar corona is radiated in optical - ir photons by the same coherent mechanism .
the corona luminosity is simply proportional to the net current flowing through the magnetosphere , and is determined by the competition between occasional twisting ( due to starquakes ) and gradual dissipative untwisting of the magnetosphere . during periods of high activity ,
when starquakes occur frequently , the magnetospheric twist may grow to the point of a global instability : when a critical @xmath177 is achieved , the magnetosphere suddenly relaxes to a smaller twist angle .
a huge release of energy must then occur , producing a giant flare ; a model of how this can happen is discussed in @xcite . a possible way to probe
the twist evolution is to measure the history of spindown rate @xmath178 of the magnetar .
the existing data confirm the theoretical expectations : the spin - down of sgrs was observed to accelerate months to years following periods of activity @xcite .
a similar effect was also observed following flux enhancement in axps @xcite .
this `` hysteresis '' behavior of the spindown rate may represent the delay with which the twist is spreading to the outer magnetosphere . since the density of the plasma corona is proportional to the current that flows through it , changes in the twist also lead to changes in the coronal opacity .
the density of the plasma corona is close to its minimum @xmath179 , and it is largely made of @xmath21 pairs rather than electron - ion plasma .
the opacity should increase following bursts of activity and affect the x - ray pulse profile and spectrum .
such changes , which persist for months to years , have been observed following x - ray outbursts from two sgrs @xcite . | persistent high - energy emission of magnetars is produced by a plasma corona around the neutron star , with total energy output of @xmath0 erg / s .
the corona forms as a result of occasional starquakes that twist the external magnetic field of the star and induce electric currents in the closed magnetosphere .
once twisted , the magnetosphere can not untwist immediately because of its self - induction .
the self - induction electric field lifts particles from the stellar surface , accelerates them , and initiates avalanches of pair creation in the magnetosphere .
the created plasma corona maintains the electric current demanded by @xmath1 and regulates the self - induction e.m.f . by screening
this corona persists in dynamic equilibrium : it is continually lost to the stellar surface on the light - crossing time @xmath2
s and replenished with new particles .
in essence , the twisted magnetosphere acts as an accelerator that converts the toroidal field energy to particle kinetic energy .
the voltage along the magnetic field lines is maintained near threshold for ignition of pair production , in the regime of self - organized criticality .
the voltage is found to be about @xmath3 gev which is in agreement with the observed dissipation rate @xmath0 erg / s .
the coronal particles impact the solid crust , knock out protons , and regulate the column density of the hydrostatic atmosphere of the star . the transition layer between the atmosphere and the corona is the likely source of the observed 100 kev emission from magnetars .
the corona also emits curvature radiation up to @xmath4 hz and can supply the observed ir - optical luminosity .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
it is well established that the universe is permeated with magnetic fields ( see , e.g. , * ? ? ?
* ) , and yet the origin of them is not well understood ( see * ? ? ?
* and references therein ) .
the problem of cosmic magnetism can be divided into two parts - the origin of seed fields and their amplification . in this paper
, we are concerned with the latter .
if a weak seed magnetic field is introduced into a turbulent medium , flow motions stretch field lines and amplify the field .
such turbulent amplification has been studied since 1950 s ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the amplification goes through three stages : ( 1 ) eddy motions in turbulence are fastest on the smallest scale , which is the dissipation scale .
therefore , the stretching of magnetic field lines occurs most actively at the dissipation scale first , and the magnetic energy grows exponentially .
( 2 ) the exponential growth stage ends , when the magnetic energy becomes comparable to the kinetic energy at the dissipation scale . the follow - up stage is characterized by a linear growth of magnetic energy , and a gradual shift of the peak of magnetic field spectrum to larger scales .
( 3 ) the amplification of magnetic field stops , when the total magnetic energy becomes comparable to the kinetic energy .
a final , statistically steady , saturation stage is reached .
@xcite proposed a scenario in which turbulent flow motions are induced via the cascade of the vorticity generated at cosmological shocks during the formation of the large scale structure .
based on a model of turbulent amplification , they estimated the strength of the intergalactic magnetic field ( igmf ) ; @xmath11 a few @xmath12 g in clusters , @xmath13 g around clusters , and @xmath14 ng in filaments . in their model ,
the intracluster medium is at a stage close to the saturation one .
but the igm in filaments are still at the linear growth stage .
the igmf has been observed with faraday rotation measure ( rm ) ; for instance , @xcite in clusters and @xcite outside clusters . for the igmf with @xmath15 , the mean value of rm
is expected to vanish , and the standard deviation represents the observation ( see , e.g. , * ? ? ?
hence , @xmath16 where @xmath17 , @xmath18 , and @xmath19 are in units of @xmath20 , @xmath12 g , and pc , respectively , has been used to extract the strength of the igmf .
here , @xmath21 is the root - mean - square ( rms ) strength of line - of - sight magnetic field , @xmath19 is the coherence length , and @xmath22 is the path length .
but the formula has been applied without a clear definition of the coherence length . here , we study length scales of magnetic field in magnetohydrodynamic ( mhd ) turbulence with very weak or zero mean magnetic field , intended for application to the igmf .
several characteristic lengths , which have been introduced in previous works , are considered .
characteristic lengths of initial seed fields in the early universe could be either very large or very small , depending on how the seed fields were generated .
however , the field structure at the linear growth and saturation stages is independent of the seed fields as long as they are sufficiently weak ( see , e.g. , * ? ? ? * ) .
therefore , the characteristic lengths should be determined by the energy injection scale ( or the outer scale ) only . and based on a scaling argument , @xcite stated that in the linear growth stage , the scale at which the stretching is most active grows as @xmath23 . in this paper , through numerical simulations ,
we investigate the growth and saturation of characteristic lengths .
then , we show how the coherence length in the rm due to the igmf is defined in terms of the characteristic lengths , and present a formula for @xmath7 . and
based on the model of @xcite , we estimate @xmath7 for clusters and filaments .
we solved the incompressible mhd equations in a periodic box of size @xmath24 using a pseudo - spectral code .
we drove turbulence in fourier space .
the forcing function has the following form : @xmath25 , where @xmath26 is a complex vector that is perpendicular to the wavevector @xmath27 .
the 22 forcing components are nearly isotropically distributed in the range @xmath28 , where @xmath29 .
the phase of each forcing component randomly fluctuates , but it has a correlation time of approximately unity .
the amplitude of each forcing component also randomly fluctuates . on average ,
each forcing component injects similar amount of energy .
( 12 components ) , 3 ( 3 components ) , and @xmath30 ( 4 components ) .
therefore the peak of energy injection occurs at @xmath31 , and the energy injection scale is @xmath32 .
a similar discrete sampling of forcing can be found in @xcite , where they used 13 components . ] in physical space this forcing @xmath33 corresponds to statistically homogeneous driving on large scales .
only the solenoidal component of the velocity field was driven .
the strength of energy injection was tuned , so without magnetic field the rms velocity becomes unity , @xmath34 , at saturation . in this representation , @xmath35 can be viewed as the velocity measured in units of the rms velocity of the system .
the density is unity and the magnetic field is multiplied by @xmath36 in our simulations , so @xmath37 can be viewed as the alfven speed in the same units .
the magnetic field consists of the uniform background field and the fluctuating field : @xmath38 . at @xmath39 ,
the magnetic field had either weak uniform component ( when @xmath40 ) or only random components ( when @xmath41 ; run @xmath42-@xmath43 , see below ) , and the velocity had a support between @xmath44 in the wavevector space .
we considered only the case where the kinetic viscosity , @xmath45 , is equal to the magnetic diffusivity , @xmath46 .
see @xcite and @xcite for further details of simulations .
simulations are denoted with @xmath47-@xmath48 , where @xmath49 or @xmath50 refers to the number of grid points in each spatial direction , @xmath51 or @xmath52 refers to hyper ( and their order ) or physical viscosity / diffusivity , and @xmath53 refers to the strength of mean magnetic field . in this paper , runs @xmath42-@xmath43 , @xmath42-@xmath54 , @xmath55-@xmath54 , and @xmath56-@xmath57 are discussed . in @xmath42-@xmath43 , the mean field was zero , and initially the spectrum of magnetic field peaked at @xmath58 .
all the runs have either very weak or zero mean magnetic field , since we intend to apply the results to turbulence in the igm , as noted in introduction .
besides the initial magnetic field configuration , runs have either different numerical resolution or different viscosity / diffusivity to explore their effects . in @xmath55-@xmath54 physical viscosity / diffusivity was used , while in others hyperviscosity / diffusion was used to extend the inertial range .
runs with @xmath59 grid points are a subset of the simulations used in @xcite . for the properties of turbulence with different initial magnetic fields , please refer an extensive study in @xcite .
figure 1 shows the time evolution of @xmath60 and @xmath61 in four simulations . here , the kinetic energy and magnetic energy densities are @xmath62 and @xmath63 .
although the four simulations have different set - ups , their time evolution looks similar .
we can clearly see three stages of magnetic energy evolution . in this limit of week or zero mean field ,
the resulting turbulence is globally almost isotropic .
* peak scale of spectrum of magnetic field * , @xmath64 : figure 2a shows the spectra of velocity and magnetic field at an epoch of saturation . in @xmath65 ,
the peak occurs at the energy injection scale .
note that the peak appears at @xmath66 rather that at 2.5 , which is a numerical artifact of discrete binning of @xmath67 . in @xmath68 ,
the peak occurs at a smaller scale .
* largest energy containing scale of magnetic field * , @xmath69 : with @xmath70 representing a power spectrum that contains equal amount of energy in each decade of @xmath67 , the peak scale of @xmath71 defines the largest energy containing scale .
figure 2a shows that for the magnetic field , the peak of @xmath72 occurs at a scale even smaller than that of @xmath68 .
* peak scale of spectrum of projected magnetic field * , @xmath73 : when a three - dimensional ( 3d ) scalar quantity is projected onto a two - dimensional ( 2d ) plane , the energy spectrum of the projected quantity becomes @xmath74 ( see , e.g. , * ? ? ?
it is a bit more complicated for vector quantities , but the qualitative behavior should be similar .
figure 2a shows that for the magnetic field , the peak of @xmath75 occurs at a scale close to that of @xmath65 .
note that the largest energy containing scale of projected magnetic field corresponds to the peak scale of @xmath76 , which is the same as the peak scale of @xmath68 .
the time evolution and saturation of @xmath64 , @xmath69 , and @xmath73 , normalized with the energy injection scale @xmath1 , are shown in figure 2b .
the rugged profiles are again a consequence of discrete binning of @xmath67 .
two points are noteworthy .
( 1 ) at saturation , @xmath77 , @xmath78 , and @xmath79 . ( 2 )
the growth pattern of those scales in the linear growth stage is not clear owing to the ruggedness in the profiles , but it seems to be between @xmath80 and @xmath4 . *
energy equipartition scale * , @xmath81 : the energy equipartition wavenumber , @xmath82 , is defined by @xmath83 the time evolution and saturation of @xmath81 are shown in figure 3a . we expect that the stretching of magnetic field lines is most active at this scale .
if then , @xmath81 should represent @xmath84 ( see introduction ) and follow @xmath4 in the linear growth stage @xcite .
indeed , @xmath81 grows with a power - law index consistent with the theoretical prediction in figure 3a . at saturation
, @xmath81 becomes close to @xmath1 .
* integral scale * , @xmath85 : the length scale of magnetic field ( and velocity too ) is often characterized with the integral scale , which is defined by @xmath86 it is well known that the integral scale has the same order of magnitude as the longitudinal and transversal integral scales , @xmath87 and @xmath88 , respectively , and in incompressible isotropic turbulence ( with reflection invariance on average ) they are related by @xmath89 ( see * ? ? ?
* curvature scale * , @xmath90 : we also consider a typical radius of curvature of field lines , @xmath90 .
we define it as the distance @xmath91 at which the average correlation drops to @xmath92 , @xmath93 where the two points at separation @xmath94 are located along the same magnetic field line and the average is taken over @xmath95 .
the factor @xmath92 is an arbitrary choice .
the time evolution and saturation of @xmath85 and @xmath90 are shown in figure 3b .
we calculated @xmath90 only for runs with @xmath59 grid points . unlike @xmath81 , @xmath85 and @xmath90
seem to grow linearly in the linear growth stage ( although the reason of the linear growth is not clear ) . at saturation , @xmath96 and @xmath97 .
hence , for instance , we model the growth and saturation of @xmath85 as @xmath98 where the eddy turnover time is defined with the vorticity around the energy injection scale at saturation as @xmath99 .
with @xmath100 for the igmf , the standard deviation of rm is @xmath101 note that in this work , the density is assumed to be constant . here
, @xmath102 is the projected , 2d magnetic field . without loss of generality , we take @xmath103 as the line of sight direction .
the projected field can be written as , as noted in section 2 .
that is , we assumed a flow of period @xmath24 in the three directions of space ( see * ? ? ?
* for further details ) . in this case
, we have a usual fourier series expansion : @xmath104 and @xmath105 , where @xmath106 is the number of grid points in each side .
here , @xmath107 , and @xmath108 denotes the coordinate of a grid point @xmath109 in the computational grid . ]
@xmath110 where we used @xmath111 .
then , the square - average of the projected field becomes @xmath112 in the last step , @xmath113 statistically in the @xmath114 plane and @xmath115 were used , and @xmath116 was substituted with @xmath117 . finally , using @xmath118 , @xmath119 so far
, we have assumed that the box size is @xmath24 . when the box size is @xmath22 ( or , when the path length is @xmath22 ) , @xmath24 s in equation ( [ eq_9 ] )
should be replaced by @xmath22 .
hence , the standard deviation of rm becomes @xmath120 since @xmath121 , the coherence length used in equation ( [ eq_1 ] ) should be given as @xmath122 .
equation ( [ eq_10 ] ) can be applied to estimate rms due to the igmf . here
, we employ the model of @xcite for turbulence and magnetic field in the igm . in clusters , turbulence is near the saturation stage with @xmath123 , where @xmath124 , the age of the universe , and @xmath125 . from equation ( [ eq_5 ] ) , then , @xmath126 . if we take the energy injection scale @xmath127 kpc , which is approximately the scale height of cluster core , @xmath128 kpc . with @xmath129 @xmath20 , @xmath130 a few @xmath12 g , and the path length @xmath131 mpc , we get @xmath132 rad m@xmath10 for clusters , which agrees with the observed rms in clusters @xcite . in filaments , on the other hand , with @xmath133 , turbulence is expected to be still in the linear growth stage , and @xmath134 .
we may take the energy injection scale @xmath135 mpc , which is the typical thickness of filaments .
it is also the typical radius of curvature of cosmological shocks in filaments @xcite , which would be the major sources to drive turbulence there .
the power spectrum of @xmath136 , the curl component of flow motions which satisfies the relation @xmath137 , in the large scale structure of the universe peaks around 5 mpc too @xcite
. then , @xmath138 kpc .
the magnetic field strength in filaments quoted in @xcite is @xmath139 ng .
but we note that the value depends on how it is averaged . with the data of @xcite ,
@xmath140 a few @xmath141 ng , @xmath142 g , and @xmath143 a few @xmath144 g , in the warm - hot intergalactic medium ( whim ) with @xmath145 k which mostly composes filaments .
the average value of @xmath146 should be most relevant to rm .
then , @xmath7 for filaments , if they intersect the line of sight with right angle , would be @xmath147 normally filaments would not intersect the line of sight with right angle .
smaller angles result in larger path lengths and so larger @xmath7 , and then the typical value of @xmath7 for filaments could be a few rad m@xmath10 .
we note that the values of @xmath148rm@xmath148 toward the hercules and perseus - pisces superclusters reported in @xcite is an order of magnitude larger than the above value , and @xcite quoted the path length , @xmath22 , which is about two orders of magnitude larger .
our findings are summarized : ( 1 ) we studied different characteristic scales of magnetic field in mhd turbulence with very weak or zero mean magnetic field .
they saturate at @xmath149 , where @xmath1 is the energy injection scale .
( 2 ) during the linear growth stage of magnetic energy , the energy equipartition scale follows @xmath150 , while the integral scale follows @xmath151 . ( 3 ) the integral scale ( actually @xmath152 ) is the relevant scale for rm .
we obtained a new formula for the standard deviation of rm , @xmath7 ( see eq .
[ [ eq_10 ] ] ) . ( 4 ) we estimated @xmath7 for filaments as well as for clusters of galaxies . finally , we note that our findings are based on a small number of incompressible numerical simulations .
they appear to be rather insensitive to numerical resolution and forms of dissipation .
however , when different initial conditions or different types of forcing are used , some of the results may be different , a question which should be investigated in the future .
we thank the anonymous referee for useful suggestions .
jc was supported in part by korea research foundation ( krf-2006 - 331-c00136 ) .
dr was supported in part by korea science and engineering foundation ( r01 - 2007 - 000 - 20196 - 0 ) .
jc and dr were also supported by korea foundation for international cooperation of science and technology ( k20702020016 - 07e0200 - 01610 ) . -1 cm and @xmath61 .
time is given both in units of the eddy turnover time defined with the vorticity around the energy injection scale at saturation , @xmath99 , ( bottom ) as in @xcite and in units of the energy injection scale divided by @xmath153 at saturation ( top ) as in @xcite.,title="fig : " ] -1 cm 0.3 cm , and magnetic field , @xmath68 , at @xmath154 .
the dissipation spectrum is also shown .
although hyper - dissipation affects the spectra near the dissipation scale , it does not strongly influence the shapes of the spectra at small wavenumbers .
right panel : time evolution of peak scales of @xmath68 , @xmath72 , and @xmath75 .
@xmath155 s are the scales normalized with @xmath1.,title="fig : " ] , and magnetic field , @xmath68 , at @xmath154 .
the dissipation spectrum is also shown .
although hyper - dissipation affects the spectra near the dissipation scale , it does not strongly influence the shapes of the spectra at small wavenumbers .
right panel : time evolution of peak scales of @xmath68 , @xmath72 , and @xmath75 .
@xmath155 s are the scales normalized with @xmath1.,title="fig : " ] | in the framework of turbulence dynamo , flow motions amplify a weak seed magnetic field through the stretching of field lines .
although the amplification process has been a topic of active research , less attention has been paid to the length scales of magnetic field . in this paper
, we described a numerical study on characteristic lengths of magnetic field in magnetohydrodynamic turbulence .
we considered the case of very weak or zero mean magnetic field , which is applicable to the turbulence in the intergalactic space .
our findings are as follows .
( 1 ) at saturation , the peak of magnetic field spectrum occurs at @xmath0 , where @xmath1 is the energy injection scale , while the most energy containing scale is @xmath2 .
the peak scale of spectrum of projected , two - dimensional field is @xmath3 .
( 2 ) during the stage of magnetic field amplification , the energy equipartition scale shows a power - law increase of @xmath4 , while the integral and curvature scales show a linear increase .
the equipartition , integral , and curvature scales saturate at @xmath3 , @xmath5 , and @xmath6 , respectively .
( 3 ) the coherence length of magnetic field defined in the faraday rotation measure ( rm ) due to the intergalactic magnetic field ( igmf ) is related to the integral scale .
we presented a formula that expresses the standard deviation of rm , @xmath7 , in terms of the integral scale and rms strength of the igmf , and estimated that @xmath7 would be @xmath8 and @xmath9 a few rad m@xmath10 for clusters and filaments , respectively . |
our data consists of 19 pointed observation of a2218 , taken with the 15 micron filter of the irc - l @xcite aboard _ akari _ @xcite .
the data was reduced using the standard irc pipeline , version 20070912 . the pipeline s median sky subtraction was utilized but resulted in dark areas , significantly around the brighter sources . to deal with this issue , and the remnants of scattered light persisting post - pipeline , a further median sky subtraction of the background areas was performed .
hot pixels were masked and removed , and the remaining bad pixels were addressed using an idl sigma filtering routine .
figure 1 shows four corresponding postage stamp sections taken from frame 19 illustrating the post - pipeline output , a median filtered mask , the median subtracted result and the sigma filtered result .
the resulting 19 images were aligned with aladin and idl s hastrom , and then average combined to give the final l15 image .
a 5@xmath0 source extraction was performed with daofind @xcite .
the resulting catalogue was eyeballed to eliminate any spurious detections , giving a final number of 565 sources .
a monte carlo completeness test was performed using an idl routine written to convolve a normalized empirical psf with the final l15 image , to create artificial sources at random positions .
the artificial sources were placed sufficiently apart from one another and known sources to avoid self - confusion .
the test was performed for 80 flux bins , covering the range of flux densities for detected sources within the l15 image . in order to reduce statistical errors
the test was repeated until an effective 18452 sources per bin was achieved .
the completeness test results show the l15 image is 50% complete to 30.7 @xmath1jy and 80% complete to 39.4 @xmath1jy .
aperture photometry was taken at random positions on the final image , excluding the edges and the brightest source , and the results were plotted as a histogram of flux density against number , see figure 2 .
the resulting distribution has an asymmetric tail that signifies the contributions from bright sources .
the combined confusion and detector noise can be represented by fitting a gaussian to the histogram , with a standard deviation of 8.33 @xmath1jy giving a 5@xmath0 sensitivity estimate of 41.67 @xmath1jy .
the mean rms per pixel of the l15 noise map is 2.20 @xmath1jy .
, scaledwidth=40.0% ]
multi waveband data of a2218 was used to identify counterparts of the l15 5@xmath0 source catalogue .
hst wfp2 f450 , f606 and f814 , palomar 200inch hale _ u , v , b , i _ and ingrid wfc _ ks _ and _ j _ images of a2218 were provided by ian smail .
spitzer irac ch 1 to 4 data was obtained via leopard and combined with mopex .
an _ akari _
s11 image was provided by myungshin i m and jongwan ko , and a spitzer mips 24 @xmath1 m image was provided by eiichi egami . the counterparting procedure identified an extra 368 sources and 3 spurious 5@xmath0 detections , giving a combined counterpart catalogue of 930 sources .
aperture photometry of the 5@xmath0 source catalogue was taken with phot @xcite using an aperture of 5.96@xmath2 and a sky annulus of radii 19.07@xmath2 and 31.0@xmath2 .
an aperture correction of 1.30 , derived using a growth curve correction method , was applied and the irc data user s manual version 1.4 @xcite conversion factor of 1.691 was used to convert from adu to @xmath1jy . for the hst images
, photometry was obtained from the published catalogue of @xcite via the ned database .
the remaining images were subject to a growth curve correction method to obtain corrected aperture photometry , using a routine written in idl . for each image one or two mean growth curves were empirically constructed .
these curves were used to calculate aperture radius and correction for each remaining source .
idl s aper was used to take the subsequent aperture photometry .
our _ ks _ band photometry was compared , where available , to previously published _
band photometry @xcite and showed a less than 2% difference .
we used eazy @xcite to gain photometric redshift estimates for our 5@xmath0 source catalogue .
eazy utilizes a minimum @xmath3@xmath4 sed fitting method , which is suitable for data sets with no available spectroscopic redshifts ( zspec ) or a biased set of zspec . in our case
the majority of the small zspec available are biased at the cluster distance .
the theoretical sed templates , used by eazy , are based on semi - analytical models , and a linear combination of templates can be fitted simultaneously .
eazy gives the option to apply priors , aimed at breaking the template colour degeneracies seen with increasing redshift .
a comparison of the resulting redshift estimates for our sources with known zspec shows that applying priors gives an improved correlation of approximately 10% .
our spectra were also fitted using the photometric redshift code illustrated in @xcite , photz from here on .
this code is uniquely optimised for fitting mid - to - far - infrared pah and silicate features seen in starburst seds .
starburst template @xcite and agn template @xcite components were simultaneously fitted by photz .
a comparison of the photz and eazy redshifts estimates for sources with prominent mid - to - far - infrared features shows a correlation of around 0.8 .
differential number counts ( dn / ds ) , corrected for incompleteness , were taken for our 5@xmath0 catalogue and normalized to a euclidean slope .
these counts were corrected for flux amplification by applying magnification factors obtained via lenstool @xcite .
figure 3 compares our corrected and uncorrected counts with previously published differential number counts .
our counts corrected for lensing show an upturn around 2 mjy and peak around 0.4 mjy , in agreement with previous counts ( e.g. , * ? ? ?
* ) and the @xcite model .
brammer , r. g. , van dokkum , p. g. , & coppi , p. 2008 , apj , 686 , 1503 efstathiou , a. , & rowan - robinson , m. 1995 , mnras , 273 , 649 elbaz , d. , cesarsky , c. j. , fadda , d. , et al . 1999 ,
a&a , 351 , l37 jullo , e. , kneib , j .- p . ,
limousin , m. , elasdttir , . ,
marshall , p. , & verdugo , t. 2007 , njph , 9 , 447 lorente , r. , onaka , t. , ita , y. , ohyama , y. , tanab , t. , pearson , c. p. 2008 , akari irc data user manual version 1.4 murakami , h. , et al .
2007 , pasj , 59 , s369 , negrello , m. , et al .
2009 , mnras , 394 , 375 onaka , t. et al .
2007 , pasj , 59 , s401 pearson , c. p. , et al .
2007 , advances in space research , 40 , 605 pearson , c. 2009 , in preparation smail , i. , kuntschner , h. , kodama , t. , et al .
2001 , mnras 323 , 839 stetson , p. b. , 1987 pasp , 99 , 191 takagi , t. , arimoto , n. , hanami , h. 2003 , mnras , 340 , 813 | we present photometry , photometric redshifts and extra galactic number counts for ultra deep 15 micron mapping of the gravitational lensing cluster abell 2218 ( a2218 ) , which is the deepest image taken by any facility at this wavelength .
this data resolves the cosmic infrared background ( cirb ) beyond the 80% that blank field
_ akari _
surveys aim to achieve . to gain an understanding of galaxy formation and evolution over the age of the universe a necessary step is to fully resolve the cirb , which represents the dust - shrouded cosmic star formation history .
observing through a2218 gives magnifications of up to a factor of 10 , thus allowing the sampling of a more representative spread of high redshift galaxies , which comprise the bulk of the cirb .
19 pointed observations were taken by _ akari _
s irc mir - l channel , and a final combined image with an area of 122.3 square arcminutes and effective integration time of 8460 seconds was achieved .
the 5@xmath0 sensitivity limit is estimated at 41.7 @xmath1jy .
an initial 5@xmath0 catalogue of 565 sources was extracted giving 39 beams per source , which shows the image is confusion limited .
our 15 micron number counts show strong evolution consistent with galaxy evolution models that incorporate downsizing in star formation . |
polars ( am her type systems ) are cataclysmic binaries in which a magnetic white dwarf accretes matter from a red dwarf star filling its roche lobe .
the accretion occurs along the magnetic field lines down to the magnetic poles of the white dwarf . the x - rays and the optical polarized flux , both originating close to the surface of the white dwarf
are modulated at the orbital period which proves the synchronism of the white dwarf rotation with the orbital motion .
a comprehensive review of these objects has been done by cropper ( 1990 ) .
+ in the case of by cam , the situation concerning the different periods in the system is not clear .
the x - ray source h0538 + 608 ( by cam ) has been identified as a polar by remillard et al.(1986 ) , on the basis of the detection of a circularly polarized optical flux .
this source shows two brightness states ( szkody et al .
1990 ) , such a behaviour being shared by most polars .
however this object is atypical by several aspects . in the uv it reveals an abnormal emission line spectrum with an enhanced nv line and a weak civ line ( bonnet - bidaud and mouchet 1987 ) .
this could be linked to the chemical composition of a secondary whose outer layers have been lost during the evolution of the system ( mouchet et al .
this is also reminiscent of what is found in some novae , suggesting a possible unnoticed nova - like event ( bonnet - bidaud and mouchet 1987 ) .
however no nova outburst is recorded in archive plates ( silber et al . 1992 ( hereafter sbior ) ) .
noteworthily , x - ray spectra of by cam obtained with the bbxrt experiment revealed an oxygen absorption edge near 0.6 kev , which intensity either requires an overabundance of oxygen or partially ionized material ( kallman et al . 1993 ) . + [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] contemporaneous x - ray and optical observations ( ishida et al .
1991 , sbior ) have revealed the presence of two very close periods around 3.3h , suggesting the asynchronism of the system ( sbior ) .
this was the second polar sharing this peculiarity : the first one being v1500 cyg , a nova which is expected to become synchronous again in less than 200 years ( schmidt and stockman 1991 , katz 1991 ) .
a third one , rx j1940.1 - 1025 , has been added recently , but contrary to both previous ones , the spin period is slightly longer than the orbital period ( friedrich et al .
1996 ) . from a compilation of previous optical polarimetric and photometric measurements of by cam , combined with new observations , piirola et al .
( 1994 ) obtained a new determination of the shortest period ( 3.3308h ) which differs significantly from the original determination ( mason et al .
1989 , hereafter mls ) .
the presence of two close periods in this system led to search for a longer period of the order of fourteen days which would be the beat period .
a large set of photometric data collected over a period of 66 days , seem to reveal a period of seven days ( silber 1995 ) , but a beat period of 14 days has also been suggested by mason et al .
( 1995a , b ) .
+ recently , time - resolved uv spectroscopy revealed a modulation of the line fluxes and of the radial velocities with the longest period identified as the orbital one ( zucker et al .
1995 ) , implying an origin far from the accretion column , contrary to the common idea that the high ionization uv resonance lines are formed close to the white dwarf . in the optical emission lines are very complex and can show up to four components ( mls ) .
+ in this paper we present an analysis of high temporal resolution spectroscopic and photometric optical data obtained at three different dates .
partial results have been presented by bonnet - bidaud et al .
the optical emission lines , composed of components arising from different regions , are good tools to constrain the different periods present in this system and to trace the geometry of the accreting flow . the optical results are compared to the uv spectroscopic results . a critical review of the different period determinations is also presented in section 5 and their interpretation is discussed in section 6 in terms of phase - drifts introduced by an asynchronous rotating magnetosphere .
= 8.8 cm = 8.8 cm spectroscopic and photometric data were obtained simultaneously at the as sao 6 m telescope in zelenchuk ( russia ) in february 1990 and march 1991 .
the observations were performed using the sp-124 spectrograph and the nef photometer at the nasmyth secondary focus of the 6 meter bolshoi azimuthal telescope ( ioannisiani et al .
1982 , vikuliev et al . 1991 ) .
the spectrograph was equipped with a 1200 lines / mm grating .
a television scanner with two lines of 1024 channels was used to record the sky and source spectra simultaneously in a photon - counting mode ( see somova et al .
( 1982 ) , drabek et al .
1986 , afanasiev et al .
1991 , for a detailed description of the instrumentation ) .
the intrinsic temporal resolution is 0.033 s. the wavelength range was 3900 - 4920 ,
4030 - 5050 and 3950 - 4970 respectively in feb . 1990 , on march 10 and march 12 1991 .
a wavelength calibration has been applied using a he - ar - ne lamp .
the aperture is a circular slit which diameter was chosen between 2 and 3 `` , according to the seeing .
the corresponding spectral resolution is @xmath12.5 .
the sky spectrum has been subtracted but no flux calibration has been applied .
the log of the observations is reported in table 1 .
photometric data were obtained simultaneously with the spectroscopic data by splitting the light beam , with about 50% of the light being sent into a 12 '' aperture one - channel fast photometer .
continuous johnson b data were recorded with a 0.1s resolution .
ubvr were also acquired in one or two occasions during each observation and have been used to compute the optical magnitude of the source using star f from 3c147 field ( neizvestny 1995 ) as a reference star .
the v magnitude values , reported in table 1 , show the source to be in a high state during the three observations , distinct from the low state observed by szkody et al .
( 1990 ) in january 1989 .
when two measurements have been done during the same night , their different values are consistent with the amplitude of the light curve ( remillard et al .
1986 , sbior ) .
+ throughout this paper , heliocentric phases refer to the ephemeris given by mls that is used as a comparison ephemeris . given the uncertainty in the determination of the different periods of the system , the accuracy given in this ephemeris is purely formal and the period ( 3.322171@xmath2h ) is used here only as a trial period to facilitate the comparison with other works ( see par .
in figure 1 we show the b light curves obtained simultaneously with the spectroscopic data , at a temporal resolution of 12.8@xmath3s for the three observing dates .
they reveal two levels of activity . on february 25 1990 and on march 12 1991
the large amplitude of the modulation is reminiscent of the flaring states observed in the optical as well as in the x - rays by ishida et al .
( 1991 ) and sbior , while on march 10 , the light curve shows a lower intensity which indicates a level closer to a pulsing state ( ishida et al .
1991 , sbior ) . for both february 25 and march 12 observations ,
the lowest levels of the light curves which occur at the end and at the beginning of the observations respectively , are consistent with the level detected on march 10 .
+ the higher fluxes in 25 feb 1990 and 12 march 1991 compared to march 10 suggest an additional source of light . during a flaring state , sbior have reported a one - peaked periodic light curve , which exhibits an increase by a factor @xmath12 between the minimum and the maximum . the feb . and march 12 curves also exhibit a total variation by a factor 2 - 2.5 , but their strong variability excludes a regular quasi - sinusoidal shape
. however , silber ( 1995 ) reports flaring state light curves which also depart from a regular shape .
+ the march 10 light curve varies in amplitude by a lower factor of 1.5 , quite consistent with the pulsing state light curves shown in sbior .
there is no clear indication of the presence of two maxima , while these bumps are usually seen in the low - level optical curves with possible unequal intensities and separated by a variable phase extension ( sbior , silber 1995 ) .
+ apart from the large amplitude main modulation , these optical light curves show shorter timescale variability . we note a significant ( @xmath1 5 min . )
dip seen near phase 0.1 on march 12 . in figure 2 , are reported two blow - ups of march 10 and march 12 light curves exhibiting strong flaring on a typical timescale of 1 - 2 minutes on march 10 and of 3 - 5 minutes on march 12 .
the fft power spectrum analysis of the data does not show significant power excess at these periods indicating that these oscillations are not coherent . on march 12 ,
quasi - periodic oscillations of the order of 30 minutes are visible in the second part of the light curve ( fig.1 ) , similar to what reported previously by sbior .
spectra averaged over the full observation have been produced for the three dates .
they are similar to previously published spectra , exhibiting the usual strong balmer and helium lines ( remillard et al .
1986 , mls ) . in order to study the orbital and rotational modulation of the line profiles ,
the spectra have been co - added with a 500s resolution which is the best compromise between a high temporal resolution and a good signal - to - noise ratio .
this results in 16 , 29 and 22 spectra respectively on 1990 feb . 25 , 1991 march 10 and march 12 .
the spectra reveal the presence of several variable components in both balmer and he lines . in fig .
3 the heii and h@xmath4 profiles normalized to the continuum are shown for the three epochs in order of increasing phase from bottom to top . large variability with phase at a given epoch as well as strong changes between the three observations
are clearly seen .
remarkably , the heii profile differs from the h@xmath4 profile in most spectra .
the most complex profiles appear in feb . 90 in both balmer and heii line while in 1991 , they are restricted to the balmer components only , the heii line showing a much more regular phase variation . + the very complex and variable line profiles observed in by cam make the radial velocity ( rv ) analysis very difficult .
in addition , blends with close weaker lines might occur ( f.i .
hei 4713 in the red wing of heii 4686 ) and affect the measurements . to isolate the different components of the complex emission lines , the profiles have been fitted with the sum of two or three gaussians of variable widths , intensities and positions , using the program spectre developed by d. pelat at meudon observatory . in this program free parameters
are the width , intensity and centre of the gaussians as well as the adjacent continuum which is fitted with a polynomial of degree one .
+ ccccccc & & & & & & + + date & line & component & fwhm & @xmath5-vel@xmath6 & k amplitude & @xmath7 + & & & km s@xmath8 & km s@xmath8 & km s@xmath8 & + & & & & & & + 25 feb .
90&heii & narrow & 200@xmath090 & 49@xmath016 & 265@xmath024 & [email protected] + & & broad & 1135@xmath0 135 & -50@xmath017 & 160@xmath024 & [email protected] + & & high velocity & 405@xmath070 & -818@xmath015 & 158@xmath028 & [email protected] + & & & & & & + 10 march 91&h@xmath4 & narrow & 165@xmath035 & 18@xmath035 & 230@xmath023 & [email protected] + & heii & intermediate width & 395@xmath0100 & -105@xmath010 & 172@xmath014 & [email protected] + & & broad & 1060@xmath0260 & -109@xmath010 & 357@xmath013 & [email protected] + & & & & & & + 12 march 91&h@xmath4 & narrow & 250@xmath045 & -93@xmath051 & 214@xmath049 & [email protected] + & heii & broad & 1110@xmath0220 & -34@xmath010 & 250@xmath016 & [email protected] + & & & & & & + + @xmath6 these values should be regarded with caution due to the large uncertainty of the absolute wavelength calibration = 15.cm the results of the gaussian fits of the heii 4686 and h@xmath4 profiles have been used to derive the velocity curves of the different components .
the measured radial velocities for heii and for h@xmath4 were fitted with a sinusoid @xmath9 , @xmath10 being the phase computed from the mls ephemeris .
the best fit parameters are reported in table 2 where the error bars correspond to a @xmath11 deviation .
the uncertainties given in table 2 do not take into account possible systematic uncertainties in the absolute calibration .
the errors in the absolute @xmath5-velocity are therefore underestimated .
+ multiple components are clearly present and the deconvolution may not be unique in some cases .
the fwhm of the different components are given in table 2 , corrected for the instrumental response . in february 90 ,
the heii line is wide and exhibits a complex profile : in addition to a narrow width ( fwhm@xmath1200 km s@xmath8 ) component and a broad ( fwhm@xmath11140 km s@xmath8 ) one , a very blue high velocity shoulder is detectable mostly between phases 0.4 and 0.6 .
the rv measurements of the narrow and broad components can be satisfactorily fitted with a sinusoidal curve , while the measurements of the blue high velocity component are more scattered ( fig .
4a , b , c ) .
the narrow and broad components appear in phase but with different amplitudes .
the high velocity component lags the two others by 0.4 in phase .
it appears not to be similar to the very high velocity component detected by mls .
indeed there is no clear indication of a red counterpart at opposite phases and the observed modulation is not compatible with an amplitude as large as @xmath1 800 km s@xmath8 found by mls .
its low amplitude and its large negative @xmath5-velocity seem to favour an origin close to the white dwarf , such as a zeeman component which has been already proposed by mccarthy et al .
( 1986 ) to explain the stationary high velocity narrow component found in the hei lines of qq vul .
at the same epoch , the balmer lines show more complex profiles than the heii ( fig .
3 ) and sinusoidal fits can not be quoted unambiguously .
+ in march 1991 , the balmer and heii lines are strongly variable .
the heii line essentially exhibits two components , one broad and one of intermediate width ( fwhm@xmath1400 km s@xmath8 ) with no evidence of the narrower ( fwhm@xmath12 200 km s@xmath8 ) component seen in feb .
90 . on march 10 ,
radial velocities of the intermediate width heii component are satisfactorily described with a sinusoidal fit , while on march 12 , this component is definitely present but its rv curve departs from a sinusoidal curve , being quasi - stationary between phase 0.6 - 1.05 . for both dates ,
the radial velocities of the broad component in heii have been fitted , though on march 12 they show a significant distortion around phases 0.2 - 0.4 ( see fig .
4f ) . the resulting amplitude is lower by 25% on march 12 than on march 10 and the radial velocities on march 10 lag the ones on march 12 by 0.11 in phase ( fig .
+ the rv measurements of the broad component in h@xmath4 also strongly depart from a well - defined sinusoidal curve , particularly on march 12 ( fig .
the distortions present in these rv curves are characterized by a shoulder present at phases 0.2 - 0.5 , at the two dates , being markedly pronounced on march 12 .
it is reminiscent of what has been seen in the flaring state data presented by sbior .
the rv curves can not be described by a sinusoidal curve in this case .
it may indicate either the presence of an additional redder component or a partial eclipse of the broad component at these phases .
noteworthily , similar deviations have been reported for v1500 cyg by kaluzny and chlebowski ( 1988 ) who have attributed them to the contribution of a second accreting column .
+ at the difference of the heii line , the h@xmath4 and h@xmath13 profiles show the presence of a narrow ( fwhm@xmath1150 - 250 km s@xmath8 ) component .
the peak is clearly observed around phases 0.5 to 0.9 and repeatable from cycle to cycle .
the velocity curve of this component is shown in figure 5 .
this component being clearly detectable at specific phases only , the corresponding fits are thus less constrained .
the amplitude are similar but an obvious lag in phase is observed for the two observations separated by two days when data are folded with the trial period ( @xmath14h ) derived by mls . +
this phase difference ( @xmath15 ) is consistent with an orbital period ( @xmath16% ) longer than the trial period . the best period derived from our narrow - line observations
only will be ( @xmath17h ) .
this is slightly longer , but consistent within the error bars , with the value derived by sbior from the h@xmath18 narrow line velocity .
this provides an independent confirmation that the period deduced from the narrow - line measurements , usually associated with the orbital period , is significantly longer than the period derived from polarimetry .
we investigate more about the nature of the two periods in section 5 .
+ a broad component is obviously present in all data , but its characteristics and its corresponding sinusoidal parameters can only be safely derived for the heii component .
its width and k amplitude are typical of those measured in other polars and the origin of this component is attributed to regions down to the accretion column ( ferrario et al .
1989 , mouchet 1993b ) . in 1991
its relative phasing with respect to the narrow h@xmath4 component is 0.42 and 0.06 on march 10 and march 12 respectively , confirming the asynchronism of the system .
+ = 8.8 cm
the true location of the emission line regions in polars is still a subject of discussion .
two main components are usually considered : a broad line associated with the accretion column and linked to the rotation of the white dwarf and a narrow line associated with the heated face of the secondary , linked to the orbital period . in the case of the complex multi - component profiles observed in by cam ,
the situation is more intricate and several other regions may contribute to build the resulting profile . in an effort to reproduce the non - standard observed components , we have investigated a region which has not been yet considered in detail , i.e. the horizontal stream of matter that links the secondary to the accretion capture region , roughly defined by the so - called magnetic capture radius .
+ = 8.8 cm before being captured by the magnetic field lines , material forms a stream of matter in the orbital plane ( liebert & stockman 1985 ) ( referred below as the horizontal stream , as opposed to the accretion column which refers to the out - of - plane part of the accreting flow ) .
this stream has been suggested as a possible contributor to the emission line profiles ( mukai 1988 , mouchet 1993a ) .
the profile characteristics of a component formed in the horizontal stream mainly depend on the position and on the extension of the emitting region .
we have simulated the emission from the stream , assuming that its path is given by the ballistic trajectory as computed by lubow & shu ( 1975 ) .
this stream is supposed to extend down to a distance @xmath19 at the capture point where the ram pressure in the horizontal stream equals the magnetic pressure ( see more details in bonnet - bidaud et al .
1996 ) . to evaluate @xmath19 ,
we have supposed that the lateral extension of the stream is constant , being taken equal to @xmath20 cm ( lubow & shu 1975 , mukai 1988 ) .
we assume a 1 m@xmath21 white dwarf mass , a typical accretion rate of 10@xmath22 g s@xmath8 , and a dipole geometry with a typical polar field of 40 mg consistent with the evaluations by cropper et al.(1989 ) and piirola et al.(1994 ) .
note however that mason et al.(1995b ) suggest evidence of a more complex field geometry with accretion onto a multipole magnetic field . for the line profile simulation ,
free parameters are the mean location of the emitting region , its extension and the emissivity of each part of the region together with the mass of the white dwarf , the inclination angle and the orbital period .
since no detailed physical model is available at present to give the exact emissivity in the lines , we made the simple assumption of an emissivity through the emitting region defined as a gaussian law with a width equal to the extension .
the emissivity is taken null outside the chosen portion of the stream .
the computed phase - resolved profiles have been convolved with a spectral resolution of @xmath23 . as illustration ,
figure 6 shows different resulting profiles computed at twenty equally spaced orbital phases for two selected emitting regions , each defined by their projection onto the line of centres and expressed in units of the distance @xmath24 between the white dwarf and the inner lagrangian point .
6a corresponds to a stream extension between 0.3 to 0.5 times @xmath24 , and fig .
6b to a region far from the white dwarf , between 0.7 and 0.9 times this distance .
an inclination angle of 70@xmath25 has been assumed .
these profiles clearly show a two - peaked orbitally modulated width ( fig .
we stress that the lines produced in the stream would therefore in general be strongly variable in shape and width , appearing alternatively narrow and broad within the same cycle .
such lines would thus be difficult to extract by the standard two - component analysis .
however such high temporal and spectral resolution measurements are difficult to derive from most data .
therefore only the phase - averaged full widths at half maximum expected from different zones of the stream have been compared to the data . in figure 7
these intrinsic widths are reported against the amplitudes k of the rv curves of the lines emitted by seven regions of variable positions ( a ) and extensions ( b ) .
the inclination angle increases along the curves from 10@xmath25 to 80@xmath25 .
as expected , the wider the emitting region is , the larger the width is .
maximum values of the order of fwhm @xmath1 400 km s@xmath8 , as observed , can only be reached for a contribution all along the stream .
high k values are obtained for the parts of the stream which are the closest to the white dwarf .
lines formed in the stream at a projected distance from the white dwarf reasonably larger than @xmath26 , the value which corresponds to a typical capture radius at @xmath121 r@xmath27 , have rv amplitudes restricted to values lower than @xmath1 800 km s@xmath8 for a reasonable ( i@xmath28 ) inclination of a non - eclipsing system .
+ for higher white dwarf masses ( 1.4 m@xmath21 ) the curves are shifted up , of the order of 10% , towards greater rv amplitudes and widths .
the plotted curves are computed for a system of a 3.33h orbital period , but the values do not increase by more than 20% for a typical shorter period of 2h . + these stream properties can be compared with the characteristics of the components that were identified in by cam .
narrow components have been found in balmer lines as well as in the heii line at different epochs ( see table 2 ) .
the properties of the narrow component measured in feb .
1990 spectra , can be satisfactorily ascribed to the stream but its width value is not much constrained .
a large region starting close to the white dwarf and a low inclination angle ( i @xmath29 ) can account for the measured values
. on march 10 1991 , the width and the amplitude of the h@xmath4 narrow line are compatible with an origin in the stream , for an inclination angle lower than 30@xmath25 and a large extension of at least 0.1 - 0.5x@xmath30 ( fig.7 ) .
the data on march 12 are marginally accounted for by an extreme case where the complete stream emits .
alternatively , the fact that this narrow component is mainly visible at phases 0.6 - 1.0 when its radial velocity passes from red - to - blue may favour instead an origin in the heated face of the secondary ( see below ) .
+ more interesting is the intermediate width component measured in the heii line on march 10 1991 .
such width is not easily produced on the secondary .
it can be produced in the stream for extended regions and high inclinations , but would have rather high velocity amplitudes .
the heii intermediate width component can not satisfy both requirements ( k amplitude and fwhm ) for an origin in the stream , the amplitude being much lower than expected from the stream ( fig.7 ) .
in addition , its relative phasing with the narrow peak seen in h@xmath4 ( 0.51 ) is far from the expected value if this last component is indeed formed in the hemisphere of the secondary .
in fact , the simulations show that the blue - to - red phasing of the stream component related to the position of the secondary is constrained to an interval of phase of 0.75 - 0.85 depending on the exact position , where phase zero refers to the inferior conjunction . + on the basis of these simulations of the line emission in the stream , we can also reconsider the possibility that the uv lines arise from such a region , as it has been suggested by zucker et al .
the mean k radial velocity amplitude of 370 km s@xmath8 for civ and of 245 km s@xmath8 for heii can be easily reproduced with an origin in the stream .
the evaluation of the uv line widths is made more difficult in the case of the wide nv , civ and siiv doublets which are not resolved with iue .
however , in the case of the heii 1640 line , the blend is made of two very close lines and does not affect the determination . using archive spectra , the average measured fwhm of the heii line is of the order of 8 .
the iue profiles along the dispersion direction being well represented by gaussian profiles of about 5.5@xmath31 fwhm at this wavelength ( cassatella et al .
after deconvolution the intrinsic width of the uv heii line is then of @xmath32 km s@xmath8 .
the long exposure time may be responsible for part of the broadening .
however such a high value is incompatible with the values expected from a stream emission ( see fig .
7 ) , unless the stream is excessively elongated .
we note that due to the de - synchronization of the system , the position of the capture radius is in fact modulated at the beat period between the spin and the orbital periods ( see section 6 ) .
we have computed this position for given co - latitude angles @xmath33 of the magnetic field and for different values of the longitudinal angle @xmath34 , varying from 0 to 360@xmath25 , to mimic the variable configuration of the magnetic field .
we find that the capture distance r@xmath35 varies very little with @xmath34 , the strongest modulation ( from @xmath121 to 27 r@xmath27 ) being found for a high inclined magnetic axis ( @xmath36 ) .
we obtain a maximum elongated stream for a minimum magnetospheric radius of @xmath37r@xmath27 , which corresponds to a projected distance on the line of the centres of @xmath10.1 x@xmath30 .
even an emission from a stream extended down to this value can not account for the widths of the lines .
an alternative origin of the uv lines in the accretion column is proposed in section 6 .
+ as suggested above , the narrow component seen in the h@xmath4 profile in march 1991 might be thought to arise from the x - ray heated hemisphere of the secondary . a lower limit of the white dwarf mass
can thus be derived using the radial velocity amplitude of this narrow component . in the case of the extreme hypothesis that the k value represents the motion of the centre of the red dwarf , values of the amplitude k between 200 and 250 km s@xmath8
can be obtained for a white dwarf mass@xmath38 0.6 m@xmath21 and an inclination angle i larger than 40@xmath25 .
if , to take into account the decentered position of the gravity centre of the illuminated hemisphere , we apply a k - correction of 1.3 , typical value derived from previous studies ( mouchet 1993a ) , the permitted range of wd masses is then 0.8 - 1.4 m@xmath21 , in agreement with the lower limit found by ishida et al . ( 1991 ) , and i should be greater than 50@xmath25 .
+ in contrast , for the uv lines , these large widths exclude a dominant origin from the heated secondary .
the heated hemisphere has been suggested by zucker et al .
( 1995 ) to contribute some fraction of the nv line .
indeed a narrow component has been clearly detected in the uv lines of am her using high resolution spectra ( raymond et al .
it has also been invoked to contribute to the uv heii line of v1500 cyg ( schmidt , liebert & stockman 1995 ) , but the formation of this line in v1500 cyg is favoured by the presence of a high temperature white dwarf which strongly contributes to the illumination of the secondary .
+ = 8.8 cm the determination of an exact ephemeris of by cam appears to be a tedious task because of the difficulty in defining adequate stable time markers in an asynchronous binary system .
mason et al .
( 1989 ) first determined a high accuracy ephemeris based on a recurrent sharp drop in the circular polarization data that was assumed to track correctly the white dwarf rotation .
they noted however that their resulting ephemeris was only of apparent high accuracy , given the highly irregular behaviour of the source and the large gaps in their data set .
strictly speaking , this first tentative ephemeris with a period of ( 3.322171@xmath39h ) can therefore only be used as a comparison ephemeris . a possible alias period near 3.351
h was noted by these authors .
+ the existence of a different longer period ( 3.373@xmath40h ) was uncovered by sbior from an analysis of the h@xmath18 narrow line velocity .
a similar value was also derived from the line intensity variations .
the narrow line emission is supposed to track the secondary motion and therefore to provide the value of the orbital period .
the 1.3% difference with mason s period was then interpreted as an evidence of a slight degree of asynchronism of the same type as observed in the classical nova - polar v1500 cygni ( stockman et al .
+ however as more data accumulate , the difference between the two periods becomes less clear .
the existence of the longer period was confirmed by more data on the h@xmath18 narrow component ( sauter 1992 , quoted in zucker et al .
1995 ) and recently by the analysis of the uv emission line velocities ( zucker et al .
1995 ) around a value ( 3.3558@xmath41 h ) , consistent with sauter s orbital period .
for the shorter period , additional polarimetric data obtained by piirola et al .
( 1994 ) were found to be consistent with a short value ( 3.3308196@xmath42h ) , largely outside the error quoted in the initial high accurate ephemeris of mls .
a possible increase of the photometric / polarimetric period has also been suggested by piirola et al . and mason et al .
( 1995a ) .
more recently , an even shorter period has been reported based on long term photometry by mason et al .
( 1995b ) and silber ( 1995 ) and has been suggested to be a combination of the spin and the orbital period .
figure 8 shows the different periods presently reported for by cam . ranging from 3.29h to 3.38h , they demonstrate the very confusing situation for this source . + in an attempt to clarify this situation , we have performed a critical re - analysis of existing data , giving particular attention to possible alias periods introduced by the large gaps between different data sets .
these gaps cause a loss of the cycle number and prevent the usual ( o - c ) analysis from being performed .
we used a minimization method in which an estimator is computed from the individual timings as : + @xmath43 ^ 2}{\sigma_{ij}^2}\ ] ] @xmath44 where p is the trial period , t@xmath45 , t@xmath46 the individual timings , @xmath47 the associated errors and n the total number of data points . the k factor is introduced to normalize the estimator to the errors .
the best period is defined at the estimator minimum .
the relevant parameters @xmath48 are not independent variables , and thus the estimator does not follow a @xmath49 distribution .
the statistical significance ( _ prob _ ) was derived from the estimator distribution computed by a monte carlo method .
we find that this method gives results comparable to the lomb periodogram ( lomb 1976 ) but has the advantage of allowing a weighting of the data .
+ using this method , we first tried to combine our optical radial velocity narrow line measurements with other measurements ( mls , sbior ) .
these four collected data points span nearly five years in total .
the best period is found around 3.38 h , with a 2-day alias at 3.16 h which corresponds to the separation of our last two points , but a number of aliases close to this best value obviously prevents a more accurate determination of the period .
+ we have also applied this method to the set of data published by piirola et al .
( 1994 ) which gathered timings of the circular polarization dips and of a few photometric broad minima . in the resulting periodogram , the period claimed by piirola et al .
is evident ( [email protected] h ) , but another longer period at a similar level of significance ( _ prob_@xmath50 ) is also present at 3.3749@xmath51 h. we note that this last period is consistent with the value derived by sbior .
a f - test performed to compare the ( o - c ) residual distributions for the two periods gives no significant difference .
the polarimetric data collected over ten years therefore reveal that both the short and long periods coexist in the source .
though at a less significant level , a short beat period at a value of [email protected] h is also present , compatible with the value derived by silber ( 1995 ) and mason et al .
( 1995b ) from long term photometry ( see section 6 ) . + the discovery of a periodicity in the uv line velocity at a period of 3.3558 h ( zucker et al .
1995 ) , close to the refined 3.3554 h long period value of the narrow h@xmath18 velocity reported by sauter ( 1992 ) , was unexpected , since it suggests an origin far from the white dwarf for the uv lines .
we have reanalysed the data of table 1 of zucker et al .
( 1995 ) . in this case
, the accuracy of the data is such that one can keep track of the cycles through the full 11 days of the observations . a standard ( o - c ) method
can therefore be used instead of a periodogram search and provide a more accurate measure of the phase variations inside the observations .
we have fitted the nv radial velocity data of the six separate observations with a sine curve at mason s period ( mls ) , used as a trial period .
the phase residuals ( o - c ) are plotted in fig .
the evident linear trend clearly confirms that mason s period does not correctly track the nv velocity .
the best period determined from the trend is p=(3.3535@xmath52h ) , in accordance with the value derived by zucker et al .
however after subtraction of this linear fit , significant residual values are clearly revealed , defining a saw - tooth shape variation through two clearly separated sets of three points .
the mid - break corresponds to a phase shift of 0.133 .
if the two sets of points had been fitted individually , they would have produced two different periods ( respectively of 3.3415@xmath53h and 3.3475@xmath54h for the first and the second sets ) , each of them shorter than the mean period derived using all points .
the different `` alias '' periods are shown in fig .
8 , together with all the other period determinations .
the period determinations in by cam appear roughly distributed in three separate groups depending on the nature of the data with long values derived from spectroscopy , intermediate values from polarimetry and x - rays light curves and short values derived from optical photometry .
but as illustrated by the uv data , there is an apparent continuum of values through the different groups ( see discussion in section 6 ) .
though it has been extensively observed , the magnetic cataclysmic binary by cam is far from been well understood .
the debate about the periods has been reactivated by the study of the uv emission lines by zucker et al .
they demonstrate the presence of the so - called orbital period in the radial velocity measurements of these lines and are led to assume a formation of the uv lines far from the white dwarf , in the orbital plane .
we have shown above that , because of their broad widths , the bulk of these lines can not be produced either in the horizontal stream or in the heated hemisphere .
the most natural contribution to the uv lines is the accreting column out of the orbital plane , as it has been previously proposed for several polars , based on the fact that their orbital variations are in phase with the broad optical components ( mukai et al .
1986 , de martino 1995 ) .
we show below that the temporal behaviour of the uv lines is indeed consistent with this origin .
+ the scattering of the periods found in by cam ( fig.8 ) can be explained by a scenario , in which , in the orbital frame , the position of the rotational axis is moving slowly with the beat period of about fourteen days . the accretion column is thus formed along different field lines according to the beat phase .
schematically , the accreted material is slowly dragged by the magnetic field during the relative motion of the white dwarf , up to a certain extent when the accretion is no more possible along these lines .
accretion has then to occur at the opposite pole , causing the jump observed in fig .
+ to compute this effect , we have calculated the position , on the white dwarf surface , of the footprints of the field lines which intercept the orbital plane at the capture radius .
this is done for a fixed given co - latitude angle @xmath33 of the dipole magnetic field and for different values of the longitudinal angle @xmath34 which measure the different phases inside the beat cycle .
when the capture region is close to the white dwarf , the accretion spot is significantly distinct from the magnetic pole and for different capture regions , its location on the white dwarf surface varies .
we find that the co - latitude @xmath55 of the footprint , with respect to the rotational axis , is weakly variable with the beat phase , while its longitude @xmath56 , defined in the orbital plane with respect to the line joining the centres of the two stars , may strongly vary along the beat cycle .
this implies a lag of the impact spot with respect to the magnetic pole and the resulting phase drift would be interpreted as an apparent period longer than the true spin period .
+ if furthermore , one assumes that the accretion occurs in the opposite hemisphere as soon as the threading point is situated at an angular distance from the magnetic axis larger than @xmath57 , then at one time , the computed longitude abruptly goes down to low values and increases again in the cycle .
we find that the sudden change of the longitude value , due to the switching of the accretion , occurs before the longitude reaches the standard value of 180@xmath25 usually expected if one assumes that the accretion occurs at the magnetic pole itself .
+ in this varying geometry , we have fully computed the phasing of the radial velocity curve produced by material falling down the magnetic lines just above the white dwarf surface .
the amplitude of the shift is determined as soon as @xmath33 and the beat period are fixed .
the curve plotted in figure 9.b ) is computed for values of @xmath58 and a beat period of 14 days , and fits the data reasonably well .
the rv phase slowly varies with the beat period as observed , distorting the period determination .
the sudden change in the phasing when the pole switches , also well reproduces the magnitude of the phase jump ( @xmath59 ) observed in the uv o - c measurements .
it appears twice during a beat cycle .
a beat value can be evaluated from the point distribution as @xmath60 days .
a pole - switching behaviour has been also suggested from photometric data by silber ( 1995 , fig.3 ) .
however large phase uncertainties are introduced , in this case , by the fact that the shapes of the light curves are variable and strongly depart from a sinusoidal curve .
+ strictly speaking , one does not expect to observe the same shape of the beat modulation for lines formed close to the white dwarf and for lines emitted at other positions along the accretion column .
the emitting region is also moving , depending on the line production mechanism , and may introduce additional drifts .
the present knowledge of the detailed emission line processes at work in these systems is not yet sufficient to allow a better evaluation of this effect .
+ an interesting consequence of this phase - drift model is the fact that any period determination is biased depending on the length of the observation .
measurements extended on more than a beat period will reveal the orbital period , while data obtained in a few days will show either a shorter period than the orbital one or will not allow any period determination if situated close to the jump .
thus a large spread of period values may result as it is indeed observed in fig .
the considerations above also apply to the broad line components usually thought to be formed in the accretion column .
interestingly , the radial velocities of the h@xmath18 broad component measured by sauter ( quoted in zucker et al . 1995 ) also show variations at the orbital period ( @xmath61 frequency ) , while they have been found to be modulated with the short spin period ( @xmath62 frequency ) by sbior , based on a set of data spread over six nights only .
we predict that the o - c measurements by sauter would mimic the same behaviour as for the nv line .
+ in addition our reanalysis of piirola data has shown that a long period is also present in the polarization flux ( see section 5.3 ) , together with an indication of the ( 2@xmath63 ) combination period .
these periods are indeed expected for a cyclotron emission produced at the basis of the accretion column ( wynn & king , 1992 ) . by combining the two periods of 3.3308h and 3.3749h found in piirola polarization data , a short ( 2@xmath63 ) period of 3.2878h
is derived independently , quite consistent with values determined by silber ( 1995 ) and mason et al .
( 1995a , b ) from photometric data . from the same two values , a long beat period ( @xmath63 ) of 10.621@xmath03@xmath64days is also derived .
this value is not strictly consistent with the range of values ( 13 - 16 days ) determined from the uv data , and with the similar 14 day period suggested by mason et al .
( 1995a ) and silber ( 1995 ) .
+ in conclusion , we have shown that the period determinations are biased depending on the temporal extension of the set of data .
the phase - drift model described above explains the inability to determine a unique adequate value for the periods , when the white dwarf is not exactly synchronized .
this simple picture has to be modified if one takes into account more physical complex configurations such as a multipole geometry ( mason et al .
1995 ) , a decentered dipole , field line distortions at the threading region ( hameury , king & lasota 1986 ) or possible inhomogeneous blobs in the infalling material .
moreover it is most probable that accretion would occur on both sides for intermediate configurations .
the orbital period value is still inaccurate .
it can , in principle , be unambigously established from the study of the absorption lines associated with the companion atmosphere .
a search for the na lines was tentatively done but without success , implying that either the companion is very faint or that it is of an earlier spectral type than had been expected ( zucker et al .
finally , the discussion of the uv line formation suffers from the absence of a @xmath5-velocity value determination , which combined with the rv amplitude should allow to constrain the emitting region in the accretion column .
this can be solved with the hubble space telescope and a higher spectral resolution than provided by the iue satellite .
+ _ acknowledgments : _ we are grateful to didier pelat for providing us his program spectre and to paul mason for useful discussions . afanasiev v.l .
, lipovetsky v.a . , mikhailov v. , nazarov e. , shapovalova a.i . , 1991 , astrofiz
sao ) , 31 , 128 bonnet - bidaud j.m .
, mouchet m. , 1987 , a&a 188 , 89 bonnet - bidaud j.m . , mouchet m. , somov n.n .
, somova t.a . , 1992 , in : stellar magnetism , eds yu.v .
glagolevskij , i.i .
romanyuk , p 186 bonnet - bidaud j.m . , mouchet m. , somova t.a . ,
somov n.n . , 1996 ,
a&a 306 , 199 cassatella a. , barbero j. , benvenuti p. , 1985
, a&a 144 , 335 cropper m. , mason k.o .
, allington - smith j.r . , et al . , 1989 ,
mnras 236 , 29p cropper m. , 1990 , space sci .
reviews , 54 , 195 drabek s.v . ,
kopylov i.m . ,
somov n.n .
, somova t.a . , 1986 ,
sao ) , 22 , 64 ferrario l. , wickramasinghe d.t . , tuohy , 1989 , apj 341 , 327 friedrich s. , staubert , r. , lamer g. et al . , 1996 , a&a 306 , 860 hameury j.m . , king a.r .
, lasota j.p . , 1986 , a&a 162 , 71 ishida m. , silber a. , bradt h.v . et al . 1991 , apj 367 , 270 ioannisiani b.k . , et al . , 1982 , in : instrumentation for astronomy with large optical telescopes , ed .
humphries , reidel , p. 3
kallman t.r .
, schlegel e.m . ,
serlemitsos p.j . , et al . , 1993 ,
apj 411 , 869 kaluzny j. , chlebowski t. , 1988 , apj 332 , 287 katz j.i . , 1991 , apj 374 , l59 liebert j. , stockman h.s . , 1985 , in : cataclysmic variables and low mass x - ray binaries , eds .
d.q . lamb & j. patterson , reidel , p 151 lomb n.r . , 1976 ,
astrophys . and space science , 39 , 447 lubow s. , shu f. , 1975 , apj 198,383 de martino d. , 1995 , in : proceedings of the cape workshop on magnetic cataclysmic variables , eds d.a.h . buckley and b. warner , a.s.p .
85 , p. 238
mccarthy p. , bowyer s. , clarke j.t .
, 1986 , apj 311 , 873 mason p.a . , liebert j. , schmidt g.d . , 1989 , apj 346 , 941 ( mls ) mason p.a .
, chanmugam g. , andronov i.l .
, et al . , 1995a , in : cataclysmic variables , eds .
a.bianchini , m. della valle & m. orio , kluwer academic publishers , p. 426
mason p.a .
, andronov i.l . ,
kolesnikov s.v . ,
pavlenko e.p .
, shakovskoy n.m . , 1995b , in : proceedings of the cape workshop on magnetic cataclysmic variables , eds .
d.a.h . buckley and b. warner , a.s.p .
85 , p. 496 mouchet m. , bonnet - bidaud j.m .
, hameury j.m . , 1991 , : 11th north american workshop on cvs and lmxbs , ed .
mauche , cambridge university press , p 247 mouchet m. , 1993a , in : proceedings of the leiscester meeting on white dwarfs , ed .
barstow , kluwer academic publishers , p 411 mouchet m. , 1993b , in : proceedings of the second haifa technion conference : cataclysmic variables and related physics , annals of the israel physical society , vol .
o. regev & g. shaviv , 208 mukai k. , 1988 , mnras 232 , 175 mukai k. , bonnet - bidaud j.m .
, charles p.a . , et al . 1986 ,
mnras 221 , 839 neizvestny s.i .
, 1995 , in : photometric systems and standard stars , ed .
moletai astronomical observatory , p. 37 piirola v. , coyne g.v . ,
takalo s.j.l . , larsson s. , vilhu o. , 1994 , a&a 283 , 163 raymond j.c . , mauche c.w . , bowyer s. , hurwitz m. , 1995 , apj 440 , 331 remillard r.a . , bradt h.v
, mcclintock j.e . , patterson j. , 1986 , apj 302 , l11 sauter l. , 1992 , ph.d .
thesis schmidt g.d . ,
stockman h.s . , 1991 , apj 371 , 749 schmidt g.d . , liebert j. , stockman h.s . , 1995 ,
apj 441 , 414 silber a.d . , 1995 , in : proceedings of the cape workshop on magnetic cataclysmic variables , eds .
d.a.h . buckley and b. warner , a.s.p .
85 , p. 302
silber a. , bradt h.v . , ishida m. , ohashi t. , remillard r.a .
, 1992 , apj 389 , 704 ( sbior ) somova t. et al . ,
1982 . in : instrumentation for astronomy with large optical telescopes , ed .
humphries , reidel , p. 283
stockman h.s . ,
schmidt g.d . ,
lamb d.q . , 1988 ,
apj 332 , 282 vikuliev n. , zinkovskij v. , levitan b. , nazarenko a. , neizvestny s. , 1991 , astrofiz . issled .
sao ) 33 , 158 szkody p. , downes r.a . , mateo m. , 1990 , pasp 102 , 1310 zucker d.b .
, raymond j.c . , silber a.d .
, et al . , 1995 , apj 449 , 310 | we present the results of fast temporal optical spectroscopy and photometry of the am her type system by cam obtained in february 1990 and march 1991 .
emission line profiles show a complex structure and are strongly variable .
the radial velocity studies of a sharp component detected in balmer lines in march confirms the non - synchronism of the system .
different possible emitting regions are discussed to explain the characteristics of the multiple line components .
it is shown from a computation of the line velocities and widths that the horizontal stream can not explain the lines of intermediate widths observed in this system .
a critical review of the different estimations of the periods found in by cam is presented .
additional periods are revealed from a re - analysis of previous optical polarimetric and uv spectroscopic data .
we show that because of the slight asynchronism between the white dwarf rotation and the orbital period , significant changes in the accretion geometry introduce a bias in the period determination depending on the length of the observations .
large phase variations are shown to exist which are well reproduced by a phase - drift model in which a magnetic dipole rotates in the orbital frame with a period of [email protected] days . |
the issue of what happens to complexity in an evolving system is of great interest . in natural ( biological ) evolution
, the naive view is that life started simple , and evolved ever more complex life forms over time , leading to that pinnacle of complexity , _
homo sapiens_. the end points of that process are of course fixed . in the beginning , life must be simple . in our present era
, there must exist intelligent organisms ( namely us ) pondering over the mystery of how we came to be .
so the _ anthropic principle _ fixes the present day as having complex lifeforms .
there is nothing within the _ modern synthesis _ of darwinism that implies a steady interpolation between these two end points .
in fact it is even plausible that more complex organisms than us existed in the past , but have since vanished into obscurity .
however , examinations of the fossil record over the phanerozoic ( the last 550 million years of the earth s history ) indicate almost no growth in complexity by a number of different measures over that period , apart from an initial large jump at the cambrian explosion.@xcite the interesting thing is to ask what one might see if looking at another evolutionary system apart from the one in which we evolved
. would we see any growth in complexity at all ? since we do nt have an extra terrestrial biology to observe ( a few martian meteorites aside ) , the only other systems available are artificial life systems evolving within a digital computer such as tierra or avida .
the avida group has reported measuring the information content ( complexity ) of individual avidan organisms@xcite , or rather a lower bound of the organism s complexity .
their results are that this lower bound increases over time for the maximally fit organism , thus showing information accumulating as time progresses
. one important critique of this work , however , is that organisms do not interact directly with each other , and in order to prevent evolution stagnating , an externally imposed task ( eg computing a logical operation ) is added to the system .
organisms are given `` fitness points '' depending on how well they perform this task .
this heavily weights the system in favour for accruing information .
by contrast , in the tierra system , the organisms interact with each other , providing a rich array of possible ( intrinsic ) tasks for the organisms to exploit .
since this is an evolving ecology with no externally imposed task , the above critique does not apply .
however , the downside is that determining whether two genotypes are phenotypically equivalent is considerably more complex . in some work a couple of years ago@xcite , i studied the phenotypic properties of tierran organisms to build up a picture of the genotype to phenotype landscape .
a tierran organism s phenotype can be characterised by a couple of numbers for each possible pairwise interaction in the ecology .
multiway interactions are ignored in this study , as experience has shown them to be relatively rare .
the information content of a string is given by the difference between the maximal shannon entropy of that string ( i.e. considering the string to be random , or devoid of information ) , and the entropy given by assuming that the string codes for some phenotype @xmath0:@xcite @xmath1 where @xmath2 is the length of the genotype ( in instructions ) , and @xmath3 is the number of genotypes that give rise to the same phenotype @xmath0 .
the base , 32 , refers to the number of instructions in the tierra instruction set . if @xmath4 ( ie a completely random sequence ) , then @xmath5 .
similarly , if @xmath6 ( there is only one genetic sequence encoding a genotype , or no redundancy ) , then @xmath7 .
the most obvious way to compute @xmath3 is to search all @xmath8 genotypes for equivalent phenotypes .
however , this is an enormous number of strings to check , and computationally infeasible .
adami recognised this problem , and took the approach of counting the number of volatile sites @xmath9 ( sites that vary amongst phenotypic equivalents ) , and approximating @xmath10 . in one sense
this is an overestimate of @xmath3 , so they argue that this gives a lower bound to the information @xmath11 . in another sense , however , it is not strictly a lower bound . if it turns out that fixing one of the volatile sites to a particular value allows one of the fixed sites to vary without altering the phenotype , then this would be not be counted in the @xmath3 .
so what we have is really an overestimate of an underestimate .
the same criticism applies to this work .
we can estimate the above mentioned estimate fairly accurately , more precisely we can find the size of the neutral network@xcite connected by one - site neutral mutations to @xmath12 .
however , the possibility remains that there are other neutral networks of @xmath12 that are nt connected by single site mutations to @xmath12 .
probably the most efficient way of finding these is by using a genetic algorithm to explore genotype space , i.e. run tierra for a long time to see what it discovers ! the way we use this in our experiment is to keep a list of neutrally equivalent organisms that tierra discovers .
as we explore the neutral network connected to @xmath12 , we eliminate items from the list that we come across .
the remaining names on the list can then be used as seeds to start the process again . in this work ,
we use two different techniques to measure @xmath3 .
the first is a monte carlo random sampling technique to estimate the proportion of the @xmath13 strings found by varying the volatile sites .
the second technique , which we use in conjunction with the monte carlo approach mentioned above , is to walk the neutral net .
the monte carlo technique works well when the density of neutral variants is fairly high , whereas the latter technique is best on sparse networks . a decision on which technique to use for
which site is based on estimated densities of neutral variants .
equation ( 4 ) of @xcite presents the dynamical equations of two species of tierran organisms interacting .
the precise form of the dynamics is not important here , however the phenotype of the organism can be characterised by its interactions will all other possible tierran phenotypes .
since it is impossible to have the complete set of all possible tierran organisms , those organisms generated during a run of tierra are used . since tierran organisms coevolve , the most important organisms should be contemporaneous with the test organism .
the following characteristics are saved for each pair of organisms : 1 . the outcome of the tournament .
this may be one of the following : + infertile : : the test organism never calls the divide instruction , or does not produce any recognisable progeny ( essentially still born ) once : : the organism produces progeny once , but then never repeats the act .
repeat : : the organism continuous reproduces the same progeny . for this purpose
we ignore what is produced first time around , as this will be swamped by number latter progeny .
nonrepeat : : the organism continuously reproduces , but the progeny is either different each time , or the cpu is in a different state each time the divide instruction is called - thus ca nt be guaranteed to reproduce ad infinitum .
the name of the progeny organism .
this is usually identical to the parent , but may another type in the case of symbiosis or parasitism .
3 . the number of timesteps it takes to reach the first divide instruction ( @xmath14 ) , and the time it takes between successive divide steps after that ( @xmath15 ) .
4 . the number of template matching operations made to the opposing organism prior to the first divide ( @xmath16 ) and between successive divides ( @xmath17 ) .
two organisms are neutrally equivalent if they have identical characteristics against all tierran organisms . once all organisms are paired with each other , we can produce a list of phenotypically unique organisms , which provides a smaller test list to pit trial mutants against .
we may also eliminate some noninteractive pairings prior to simulation by trying to see if potential template matches could happen between organisms .
this still produces a fairly large list of test organisms , so it is still computationally expensive .
the high degree of parallelism in this problem allows it to be attacked in reasonable time on a parallel supercomputer . a further refinement may be possible by producing an archetypal list , perhaps by ignoring the ( @xmath18 and @xmath19 ) parameters
the idea being that the archetypes contain a representative organism from each niche of the ecology , and ignoring minor differences such as reproductive rate .
this would coarsen the approximation a little , but will probably give an acceptable result . at present this idea has not been tested .
due to the time constraints of producing this paper , the analysis of a reasonable length tierra run has not been completed . at the time of writing , a moderately large data set of 1660 organisms
was generated from a 24 hour tierra run .
tierra produces most of its diversity during the earliest stage of its running , so it becomes significantly more expensive to produce larger data sets .
this data set was halved by removing every second organism , and then a phenotypic analysis was carried out .
this set reduced to 103 distinct phenotypes , which formed the test list used for carrying out the complexity analysis .
each of these 103 organisms were then tested for phenotypic equivalence against their single site nearest neighbours .
the number of sites on which no mutation resulted in a phenotypically equivalent organism ( `` nonvolatile sites '' ) is plotted against the time of speciation in figure [ results ] . | recently , adami and coworkers have been able to measure the information content of digital organisms living in their _ avida _ artificial life system .
they show that over time , the organisms behave like maxwell s demon , accreting information ( or complexity ) as they evolve . in _ avida
_ the organisms do nt interact with each other , merely reproduce at a particular rate ( their fitness ) , and attempt to evaluate an externally given arithmetic function in order win bonus fitness points .
measuring the information content of a digital organism is essentially a process of counting the number of genotypes that give rise to the same phenotype .
whilst avidan organisms have a particularly simple phenotype , tierran organisms interact with each other , giving rise to an ecology of phenotypes . in this paper ,
i discuss techniques for comparing pairs of tierran organisms to determine if they are phenotypically equivalent .
i then discuss a method for computing an estimate of the number of phenotypically equivalent genotypes that is more accurate than the `` hot site '' estimate used by adami s group .
finally , i report on an experimental analysis of a tierra run . |
the haldane shastry ( hs ) spin chain describes @xmath1 spins equally spaced on a circle with an interaction inversely proportional to the square of their chord distance @xcite .
the original motivation for studying this model is the fact that it possesses an exact jastrow - product ground state , which coincides with the @xmath2 limit of gutzwiller s variational wave function for the hubbard model @xcite , and also with the one - dimensional version of the resonating valence bond state introduced by anderson @xcite . since its very introduction
, the hs spin chain has been extensively studied as a completely integrable model @xcite solvable by the asymptotic bethe ansatz @xcite , whose spinon excitations provide a simple example of a system obeying fractional statistics @xcite .
the energy spectrum of the hs hamiltonian with spin @xmath3 was partially computed in the original papers of haldane and shastry . in a subsequent publication
@xcite , haldane et al . empirically found a complete description of the spectrum for arbitrary spin , and explained its highly degenerate character by the symmetry of the model under the yangian algebra @xmath4 .
these results were rigorously established in ref .
@xcite by explicitly constructing a transfer matrix in terms of the dunkl operators @xcite of the trigonometric sutherland dynamical model @xcite . in this approach ,
the spectrum is obtained by considering all possible _ motifs _
@xmath5 , where each @xmath6 is either @xmath7 or @xmath8 and the maximum number of consecutive @xmath8 s is @xmath9 .
indeed , the energy associated with a motif @xmath10 is given by the compact formula @xmath11 the degeneracy of a level @xmath12 is obtained by summing the degeneracies corresponding to all the motifs @xmath10 such that @xmath13 .
although there is a well - defined algorithm for computing the degeneracy of each motif , in practice the computation becomes quite involved except for @xmath14 .
it is therefore difficult to derive in this way an exact expression for the partition function valid for arbitrary values of @xmath1 and @xmath15 .
perhaps as a consequence of this fact , little attention has been paid in the literature to the global properties of the spectrum of the hs chain .
some authors @xcite have suggested that the main obstacle in computing the partition function of the hs chain in closed form is the fact that the dispersion relation is nonlinear in @xmath16 , in contrast with the polychronakos rational chain @xcite . in a recent paper @xcite ,
however , the partition function of the trigonometric hs spin chain of @xmath17 type has been exactly computed applying what is known as polychronakos s _ freezing trick _
@xcite , notwithstanding the fact that these chains have a nonlinear dispersion relation similar to .
in fact , we shall prove in what follows that the partition function of the chain can also be computed using the freezing trick . from the partition function
it is straightforward to generate the spectrum of the hs chain for a wide range of values of @xmath1 and @xmath15 , and thus study global properties thereof such as the level density or the distribution of the spacing between consecutive levels .
for convenience , we shall take the hamiltonian of the ( antiferromagnetic ) haldane shastry spin chain as @xmath18 where @xmath19 and @xmath20 is the spin permutation operator of particles @xmath21 and @xmath16 . here and throughout the paper all
sums and products run from @xmath8 to @xmath1 unless otherwise specified .
the hamiltonian of the original hs spin chain is given by @xmath22 , where @xmath23 is the highest energy of @xmath24 . in order to apply the freezing trick ,
we need to introduce the sutherland spin model @xmath25 and its scalar version @xmath26 we thus have @xmath27 where @xmath28 is obtained from @xmath24 by the replacement @xmath29 .
the freezing trick is based on the fact that for @xmath30 the particles `` freeze '' at the equilibrium positions of the scalar part of the potential in @xmath31 , which are simply the lattice points of the chain . in this limit ,
the spin degrees of freedom decouple from the dynamical ones , so that by eq .
the energies of the dynamical spin model are approximately given by @xcite @xmath32 where @xmath33 and @xmath34 are _ any _ two levels of @xmath35 and @xmath24 . hence the partition functions @xmath36
, @xmath37 , and @xmath38 of @xmath24 , @xmath31 , and @xmath35 , respectively , satisfy the approximate equality @xmath39 the latter equation leads to the _ exact _ formula @xmath40 which we will use to compute the partition function of the chain in closed form .
in order to evaluate the rhs of , we need to compute the spectra of @xmath31 and of its scalar limit @xmath35 .
these spectra can be obtained in a unified way by considering the scalar differential - difference operator @xmath41 where @xmath42 permutes the coordinates @xmath21 and @xmath16 .
the operator @xmath43 is represented by an upper triangular matrix in a ( non - orthonormal ) basis whose elements are of the form @xmath44 where the vector @xmath45 is such that the differences @xmath46 , @xmath47 , are integers .
the basis elements should be ordered in a suitable way that we shall now describe .
we shall say that a vector @xmath48 is _ nonincreasing _ if @xmath49 for @xmath50 . given two nonincreasing vectors @xmath51 and @xmath52
, we shall write @xmath53 if @xmath54 and @xmath55 .
finally , we say that the basis element @xmath56 precedes @xmath57 if @xmath53 , where @xmath51 and @xmath52 are the unique nonincreasing vectors obtained from @xmath58 and @xmath59 by reordering their components .
it can then be shown that the matrix of @xmath43 in the basis @xmath60 with the order just defined is indeed upper triangular , with diagonal elements @xcite @xmath61 we shall now see how the spectrum of @xmath31 follows easily from that of @xmath43 . to this end , let us introduce the total antisymmetrizer @xmath62 with respect to simultaneous permutations of the spatial and spin coordinates . we can construct a ( non - orthonormal ) basis of the hilbert space of the hamiltonian @xmath31 with states of the form @xmath63 where @xmath64 is an element of the spin basis and the vector @xmath58 satisfies the following conditions : .8cmthe differences @xmath65 , @xmath47 , are nonnegative integers . at most @xmath15 components of @xmath58 can be equal . the total momentum vanishes , i.e. , @xmath66 .
the first two conditions are a direct consequence of the antisymmetric nature of the states .
the last condition reflects the fact that , since @xmath31 is translationally invariant , we can work in the center of mass frame .
the basic states @xmath67 should be ordered in such a way that @xmath67 precedes @xmath68 if @xmath69 ( note that the vectors @xmath58 and @xmath59 are nonincreasing by condition i ) ) . from the elementary relations @xmath70 and
the fact that @xmath43 clearly commutes with @xmath62 , it follows that @xmath71 & = { \lambda}\big(\be({\mathbf{p}})\phi_{\mathbf{p}}\ket{\mathbf{s}}+\sum_{{\mathbf{p}}'\prec{\mathbf{p}}}c_{{\mathbf{p}}{\mathbf{p}}'}\phi_{{\mathbf{p}}'}\ket{\mathbf{s}}\big)\\ & = \be({\mathbf{p}})\psi_{{\mathbf{p}},{\mathbf{s}}}+\sum_{{\mathbf{p}}'\prec{\mathbf{p}}}c_{{\mathbf{p}}{\mathbf{p}}'}\psi_{{\mathbf{p}}'\!,{\mathbf{s}}}\,.\end{aligned}\ ] ] hence the hamiltonian @xmath31 of the sutherland spin model is upper triangular in the basis @xmath72 , with diagonal elements @xmath73 where @xmath58 satisfies conditions i)iii ) above .
the spectrum of @xmath35 can be derived by a similar argument , noting that @xmath74 on scalar symmetric states of the form @xmath75 , where @xmath76 is the symmetrizer with respect to the spatial coordinates and @xmath58 satisfies only conditions i ) and iii ) above .
hence @xcite the eigenvalues @xmath77 of @xmath35 are also given by the rhs of , where now @xmath58 is not restricted by condition ii ) . from the above results it is easy to compute the partition functions @xmath78 and @xmath79 in the limit @xmath30 .
for the computation of @xmath78 , we start by expanding the eigenvalues of @xmath35 in powers of @xmath80 as @xmath81 where @xmath82 since @xmath83 does not depend on @xmath58 , and therefore contributes the same overall constant factor to both @xmath38 and @xmath37 , we shall henceforth drop the first term in eq . .
with this convention , for @xmath84 the denominator in eq . is given by @xmath85 where @xmath86 and the outer sum runs over all vectors @xmath58 satisfying conditions i ) and iii ) above . setting @xmath87 we have @xmath88 taking into account that @xmath89 is determined by the remaining @xmath90 s by condition iii ) , we finally obtain @xmath91 in order to compute the partition function @xmath79 for @xmath92 , it is convenient to represent the vector @xmath58 labeling the energies of @xmath31 as @xmath93 note that @xmath94 , so that @xmath95 belongs to the set @xmath96 of partitions of @xmath1 ( taking order into account ) . calling @xmath97 and dropping again the term @xmath98 , in the large @xmath80 limit eq
. becomes @xmath99 where @xmath100 since @xmath101 does not depend on the spin coordinates @xmath102 , the degeneracy associated with this eigenvalue is given by @xmath103 so that @xmath104 if @xmath105 for some @xmath21 , in accordance with condition ii ) .
hence @xmath106k_1\rho_1+\cdots+k_r\rho_r=0}}\ ! q^{\,\sum\limits_{i=1}^r\rho_il_i}.\ ] ] calling @xmath107 , @xmath108 , and @xmath109 , we have @xmath110 where @xmath111 by eq . .
note , in particular , that the numbers @xmath112 depend on @xmath113 through the partial sums . substituting into , and taking into account that @xmath114 implies @xmath115 , we obtain @xmath116 combining eqs . and , the partition function @xmath36 can be expressed in closed form as @xmath117 note that , by definition , the partial sums @xmath118 are natural numbers satisfying @xmath119 . denoting by @xmath120 the elements of the set @xmath121 and setting @xmath122 we have @xmath123 this identity and eq . yield the following remarkable formula for the partition function of the spin chain : @xmath124 from the previous formula it follows that the energy levels of @xmath24 are of the form @xmath125 where @xmath126 if @xmath16 is one of the partial sums @xmath127 corresponding to a partition @xmath128 with @xmath129 for all @xmath130 ( by condition ii ) ) , and @xmath131 otherwise . in order to relate eq . with the known expression for the energies of the original hs hamiltonian
, we need to evaluate the maximum energy @xmath132 . from eq .
we have @xmath133 hence @xmath134 where the last sum is evaluated in ref . @xcite . since the rhs of coincides with the sum @xmath135 , eq
. implies eq . with @xmath136 . in particular , from the latter relation between @xmath10 and @xmath137
it follows that @xmath10 is a motif with no more than @xmath9 consecutive @xmath8 s .
the rhs of eq . is a polynomial in @xmath138 whose evaluation with a symbolic algebra package is straightforward once @xmath1 and @xmath15 are fixed . in this way we have been able to compute the spectrum of the chain for relatively large values of @xmath1 and @xmath15 ,
for which the usual motif approach becomes inefficient due to the difficulty of computing the degeneracies . from the analysis of the spectral data
thus obtained one can infer several global properties of the spectrum that we shall now discuss .
in the first place , it is apparent that for @xmath139 the level density is gaussian to a very high degree of accuracy , as in the hs spin chain of @xmath140 type studied in ref . @xcite . in other words , for large @xmath1
the cumulative level density @xmath141 is approximately given by @xmath142,\ ] ] where @xmath143 is the degeneracy of the energy @xmath144 , and @xmath145 and @xmath146 are respectively the mean and the standard deviation of the energy .
this can already be seen , for instance , in the case @xmath147 and @xmath14 presented in fig .
[ levden ] .
the agreement between @xmath148 and @xmath149 rapidly improves as @xmath1 and/or @xmath15 grow , e.g. , for @xmath14 the mean square error decreases from @xmath150 for @xmath147 to @xmath151 for @xmath152 , or from @xmath153 for @xmath147 to @xmath154 for @xmath152 when @xmath155 . since
, by the previous discussion , for large @xmath1 the level density is characterized by @xmath145 and @xmath146 through the gaussian law , it is of interest to compute these parameters in closed form as functions of @xmath1 and @xmath15 . in the first place , using the identity @xmath158 and eqs . and
, we obtain @xmath159 similarly , the formula @xmath160 yields @xmath161 ( cf . ref .
@xcite for the last equality ) .
the level density is also gaussian as @xmath162 for the so - called `` embedded gaussian ensemble '' ( egoe ) @xcite in random matrix theory .
note , however , that in the egoe this property is valid provided that the number of one - particle states tends to infinity faster than @xmath1 .
this additional condition clearly does not hold in our case , since the number of one - particle states ( i.e. , @xmath15 ) is fixed .
another characteristic feature of the egoe is the fact that the nearest - neighbor spacing distribution @xmath163 is approximately given by wigner s law @xmath164 as for the classical gaussian orthogonal ensemble @xcite .
on the other hand , since the hs spin chain is integrable , one would expect that its nearest - neighbor spacing distribution obey poisson s law @xmath165 , according to the conjecture of berry and tabor for a generic integrable model @xcite .
this conjecture has been verified for a variety of integrable many - body problems , such as the heisenberg chain , the @xmath166-@xmath167 model , the hubbard model @xcite , and the chiral potts model @xcite .
one of the main results of this paper is the fact that the nearest - neighbor spacing distribution of the hs chain deviates substantially from both wigner s and poisson s laws . in order to correctly take into account the effect of the local level density in the study of @xmath163
, one must first apply to the `` raw '' spectrum the so - called _ unfolding _ mapping @xcite .
this mapping is defined by decomposing the cumulative level density @xmath168 as the sum of a fluctuating part @xmath169 and a continuous part @xmath170 , which is then used to transform each energy @xmath144 , @xmath171 , into an unfolded energy @xmath172 . the function @xmath163 is defined as the density of the normalized spacings @xmath173 , where @xmath174 is the mean spacing of the unfolded energies . by the previous discussion , in our case we can take the unfolding mapping @xmath170 as the cumulative gaussian distribution @xmath157 with parameters @xmath145 and @xmath146 given by the previous formulas . as for the level density , to compare the discrete distribution function @xmath163 with a continuous distribution it is more convenient to work with the cumulative spacing distribution @xmath175 .
our computations for a wide range of values of @xmath1 and @xmath15 show that @xmath176 is essentially different from either poisson s or wigner s law , since its slope tends to infinity both as @xmath177 and @xmath178 , where @xmath179 is the largest spacing .
in fact , it turns out that in all cases @xmath176 is well approximated by a cumulative distribution of the simple form @xmath180,\ ] ] where @xmath181 and @xmath182 .
the parameter @xmath183 is fixed by requiring that the average spacing be equal to @xmath8 , with the result @xmath184 where @xmath185 is euler s beta function .
for instance , for @xmath186 and @xmath14 the largest spacing is @xmath187 , and the best least - squares fit parameters @xmath188 and @xmath189 are respectively @xmath190 and @xmath191 , with a mean square error of @xmath192 ( see fig . [ spacings ] ) . and its approximation @xmath193 ( grey line ) for @xmath186 and @xmath14 .
for convenience , we have also represented poisson s ( long dashes ) and wigner s ( short dashes ) cumulative distributions.[spacings],width=302 ] for a fixed value of @xmath15 , the parameters @xmath188 , @xmath189 and @xmath179 vary smoothly with @xmath194 , provided that @xmath1 has a fixed parity of a fixed parity , but decreases when @xmath1 jumps from @xmath195 to @xmath196 . ] .
for instance , in fig .
[ albesmax ] we plot these parameters for @xmath14 and odd @xmath1 running from @xmath197 to @xmath198 ( the plot for even @xmath1 is very similar ) . in all cases ,
the fit of the distribution to the data is quite good , the mean square error never exceeding @xmath199 .
we have performed a similar analysis for @xmath155 and @xmath200 , obtaining totally analogous results .
the divergence of the nearest - neighbor spacing distribution @xmath163 for small @xmath202 is probably related to the flatness of the tail of the gaussian distribution .
it could also be argued that , since the haldane shastry chain is completely integrable , the full spectrum is a superposition of the spectra of the hamiltonian restricted to subspaces of common eigenfunctions of a suitable family of commuting first integrals .
it is well known , in this respect , that a superposition of a large number of unrelated spectra leads to a sharp increase in the number of very small spacings @xcite . on the other hand
, we do not have a clear explanation of the fact that @xmath163 also diverges when @xmath202 approaches the largest spacing @xmath179 .
this fact , which certainly deserves further study , could be a characteristic property of all spin chains of haldane shastry type .
our results also imply that berry and tabor s conjecture does not hold for the hs spin chain , even if we restrict ourselves to a subspace of the whole hilbert space with well - defined quantum numbers .
indeed , the nearest - neighbor spacing distribution of the superposition of even a small number of spectra with poisson - distributed spacings must also be of poisson type @xcite . as an illustration of these assertions , we present in fig .
[ fixspin ] a plot of the cumulative spacing distribution corresponding to the restriction of the hamiltonian to the subspace with zero total spin and odd parity for @xmath203 and @xmath14 , obtained by a numerical computation of the spectrum of @xmath24 restricted to this subspace .
it is apparent from this plot that @xmath176 is neither poissonian nor of wigner type , and that it is well approximated by a function of the form for spacings @xmath204 .
it is also clearly noticeable that @xmath163 tends to infinity as @xmath202 approaches the maximum spacing @xmath205 .
( solid dots ) for states with zero total spin and odd parity when @xmath203 and @xmath14 . for comparison purposes ,
we have represented poisson s ( long dashes ) and wigner s ( short dashes ) cumulative distributions.[fixspin],width=302 ] the non - poissonian behavior of the spacing distribution could in principle be due to finite - size effects @xcite .
although this possibility should be explored in more detail , our data clearly show that the cumulative spacing distribution @xmath176 is of the form for a wide range of values of @xmath206 .
note , finally , that an interesting integrable model not obeying the berry
tabor conjecture has been recently constructed in ref .
in contrast with the hs spin chain , the latter model is a non - generic element of a class depending on a large number of parameters , and involves many - body interactions . | we derive an exact expression for the partition function of the @xmath0 haldane shastry spin chain , which we use to study the density of levels and the distribution of the spacing between consecutive levels .
our computations show that when the number of sites @xmath1 is large enough the level density is gaussian to a very high degree of approximation .
more surprisingly , we also find that the nearest - neighbor spacing distribution is not poissonian , so that this model departs from the typical behavior for an integrable system .
we show that the cumulative spacing distribution of the model can be well approximated by a simple functional law involving only three parameters . |
the recent accelerated expansion of the universe and its cause is one of the most puzzling mysteries in cosmology and physics today . since the observations of type ia supernovae ( sneia ) reported by two groups @xcite
, a lot of observations have been accumulating to indicate that the universe is well described by the @xmath3cdm model ( for recent observational results , see _
e.g. _ ref .
although the cosmological constant ( cc ) gives a good fit to the data and explains well the current cosmic acceleration , future observations promise to provide much tighter constraints of the models .
that motivated theorists to explore other possibilities , which generically came to be known as _ dark energy _ ( de ) ( for a review on dark energy see _
e.g. _ ref .
the origin of dark energy could be matter whose properties mimic those of cosmological constant , but it could be also in the modification of gravity on cosmological scales @xcite .
in fact , due to the intricate coupling between gravity and matter in some theories , it is not always possible to tell whether dark energy comes from a new kind of matter or from a modification of gravity . in this work
we examine the idea that dark energy originates from the backreaction of quantum fluctuations originating in the primordial inflationary universe .
the idea that the origin of dark energy can be linked to primordial inflation has not been widely explored ( for an early attempt to link the cause of inflation with the cause of dark energy see ref .
@xcite , for a more recent attempt see @xcite ) .
also , early attempts @xcite to link quantum fluctuations of the _ inflaton _ to dark energy turned out not to be correct @xcite . indeed , a careful one - loop calculation of the energy - momentum tensor from inflationary gravitons @xcite , and an educated estimate of the corresponding energy - momentum tensor from scalar cosmological perturbations @xcite shows that , rather than contributing to dark energy , these inflationary perturbations contribute a tiny amount ( about @xmath4 of the critical density today ) to dark matter .
recently , the idea of relating quantum backreaction to dark energy has again drawn some attention , and here we give a brief overview . in @xcite
the minimally coupled massless spectator scalar was studied where it was found that the quantum backreaction in late - time matter era scales just as nonrelativistic matter fluid driving the background expansion , but its fraction is tiny , about @xmath4 of the critical density .
this ratio is determined by the same ratio reached by the end of inflation , and it effectively freezes afterwards .
the same was concluded for the contribution from gravitons .
the result for scalars was subsequently confirmed in @xcite .
there it was also investigated what are the influences of possible pre - inflationary periods on the magnitude of late - time quantum backreaction , and it was found it could be increased if additional inflationary periods at a much higher ( planck ) energy scale existed prior to the standard one , but the late - time scaling can not be changed .
however , when one considers the quantum backreaction from inflationary quantum fluctuations of a massless non - minimally coupled spectator scalar field , and when the relevant non - minimal coupling is negative ( such that it gives a tachyonic mass to the scalar during inflation ) , then under reasonable conditions on inflation and non - minimal coupling , the scalar field can yield a large quantum backreaction at late times @xcite , making it potentially a candidate for dark energy ( the analysis performed in @xcite is perturbative , and hence any statements about whether that is a reliable candidate for dark energy can not be made ) . in this model
the backreaction can grow considerably during inflation ( even to nonperturbative values ) , how fast depending on the nonminimal coupling ( this was already found in @xcite for more general slow - roll inflation ) , so its ratio to the background fluid is greatly enhanced compared to the minimally coupled case .
this ratio again freezes during radiation period , but starts to evolve again in matter period .
another idea involves the late - time quantum backreaction from inflationary quantum fluctuations of a very light spectator scalar field @xcite , where the backreaction contributes like a cc at late - time matter era . in order for this idea to work , the authors of ref .
@xcite needed to lower the scale of inflation and furthermore they needed a humongously large number of e - foldings of inflation ( @xmath5 ) . in a more recent paper @xcite it was pointed out that the mechanism works as well for inflation at the grand - unified scale and that the required number of e - foldings is ` only ' about @xmath6 .
when compared with the original work @xcite , that was a significant improvement . in @xcite it
was also studied how some pre - inflationary periods can lead to lowering the requirements on the number of e - foldings of this model where they managed to get it down to @xmath7 with the assumption of another planck scale inflationary period preceding the gut scale one .
apart from various technical improvements , the goal of this paper is to study the late - time quantum backreaction from inflationary quantum fluctuations of scalar fields and its relation to dark energy in more general models , without relying on any pre - inflationary physics , and one of the important result of this work is the realization that the simple addition of a small non - minimal coupling to the ricci scalar can completely alleviate the constraint on the length of inflation . we show that the late time one - loop quantum backreaction from a nonminimally coupled , light spectator scalar field can be a good candidate for dark energy . in our model conditions on inflation are completely relaxed : inflation can occur at the grand unified scale and it can last as little as hundreds of e - foldings .
the model of a very light , nonminimally coupled spectator scalar was studied and proposed as a dark energy model very soon after the discovery of the recent accelerated expansion of the universe by parker and raval @xcite . in those works the main effect derives from the ultraviolet quantum fluctuations , as opposed to the work presented here where the main contribution to the energy - momentum tensor comes from the infrared ( super - hubble ) quantum fluctuations .
we provide a more detailed comparison of our work with that of parker and raval in section [ sec : discussion and outlook ] .
quantum fluctuations of fields are generally nonvanishing , so we expect them to contribute as corrections to the classical einstein s equations .
this statement can be neatly summarized by the following equation , @xmath8 \ , , \label{semiclassicalgr}\ ] ] which can be recognized as a quantum corrected einstein s equation the two source terms on the right are the classical energy - momentum tensor of matter fields , and the _ quantum backreaction _
, respectively .
the quantum backreaction is the expectation value of the energy - momentum tensor operator of quantum fluctuations of matter fields , and the metric field .
the evidence that quantum fluctuations indeed interact gravitationally comes from studying the fluctuations in the spectrum of the cmb , and ultimately attributing it to the spectrum of primordial inflationary quantum scalar fluctuations . as opposed to the spatial variations of the quantum fluctuations , which ultimately contribute to the cmb temperature fluctuations , here we study the homogeneous nonvanishing contribution of quantum fluctuations .
we want to study the effects of the backreaction term in cosmology , in particular its influence on the dynamics of the large scale expansion of the universe , and its possible connection to the dark energy problem .
that is a rather ambitious task , and in this work we opt to address less ambitious , but still very important questions : * can the backreaction every become large enough to influence the expansion dynamics ? * when does it become large and how it depends on the parameters of the model and the expansion history ?
* what is the behaviour of the backreaction when it becomes large and does it tend to accelerate the expansion ?
since we are interested in modeling de , we want the backreaction not to spoil the previous expansion history .
therefore , its influence on the expansion has to be negligibly small up until the recent onset of accelerated expansion .
therefore , we study equation perturbatively , in the sense that we do not consider the backreaction actually backreacting on the classical evolution , since it is small by assumption ( we thoroughly check whether this assumption is satisfied in different epochs in the history of the universe ) .
the formalism appropriate for this study is the quantum field theory in curved space - time , originating in the late 60 s and early 70 s , and by now a well established subject , covered in standard references @xcite . of course , when the backreaction becomes large , this approach breaks down , and a full self - consistent solution is needed .
here we study the backreaction from quantum fluctuations of a very light , nonminimally coupled , spectator scalar field as they evolve from an initial state specified at the beginning of inflationary period of our universe , through the radiation and matter dominated era , up until the onset of the late - time acceleration period ( see fig .
[ expansion history ] for the schematic depiction of the expansion history ) .
the leading order contribution to one - loop expectation value of the energy - momentum tensor of these quantum fluctuations is computed in each era , as a controlled expansion in small ratios of physical parameters .
this paper is organized as follows .
the following section introduces definitions and conventions for the cosmological space - time .
the third section presents the scalar field model , outlines its quantization , and defines the main quantities to be computed the scalar field mode function and the expectation value of the energy - momentum tensor . the representation and approximations of the evolution of the mode function
is given in section [ sec : evolution of mode function ] . in section [ sec : energy density and pressure integrals ]
the quantum backreaction energy density and pressure integrals are analyzed from a general point of view , and the relevant contributions are identified .
section [ sec : mode functions ] is devoted to calculating approximate mode functions on constant epsilon backgrounds , in particular their expansion in the small mass limit . in section [ sec : energy density and pressure ] the dominant contributions to energy density and pressure of the backreaction are evaluated . in the concluding section [ sec : discussion and outlook ]
we summarize the results and discuss their connection to dark energy .
an outline of the future work is also given .
the line element of a @xmath9-dimensional , spatially flat friedmann - lematre - robertson - walker ( flrw ) space - time is given by @xmath10 \
, , \ ] ] where @xmath11 is the metric , @xmath12 is the scale factor , @xmath13 denotes physical ( cosmological ) time , and @xmath14 conformal time .
the physical and conformal time are related via @xmath15 . in this work
we prefer to perform computations in conformal time , for which the metric is conformally flat , @xmath16 , and @xmath17 is the @xmath9-dimensional minkowski metric .
all the expressions are written in @xmath9 dimensions in order to facilitate dimensional regularization utilized in computing quantum expectation values ( @xmath18 is the number of physical space - time dimensions ) .
we adopt the natural units convention ( @xmath19 ) , unless explicitly stated .
the geometric conventions we use are @xmath20 for christoffel symbols , @xmath21 for the riemann tensor , @xmath22 for the ricci tensor , and @xmath23 for the ricci scalar .
the dynamics of the scale factor is governed by the friedmann equations , @xmath24 @xmath25 where @xmath26 and @xmath27 are the energy density and pressure of the @xmath28-th matter fluid , respectively , @xmath29 is the conformal hubble rate related to the physical one @xmath30 via @xmath31 , and a prime denotes differentiation with respect to conformal time .
the ( noninteracting ) matter fluids each satisfy the conservation equation , @xmath32 they are usually assumed to be ideal fluids , characterized by a linear equation of state , @xmath33 with a constant equation of state parameter @xmath34 . using this equation of state
the conservation equation can be easily integrated to yield the scaling of the fluid s energy density and pressure , @xmath35 where @xmath36 and @xmath37 the expansion history of our universe consists of a few eras in which one fluid dominates over the others that can be neglected . in such regimes
friedmann equations are readily solved for the scale factor and the hubble rate , @xmath38^{\frac{1}{\epsilon-1 } } \ , \qquad \mathcal{h}(\eta ) = \mathcal{h}_0 \bigl ( \frac{a}{a_0 } \bigr)^{1-\epsilon } \
, , \end{aligned}\ ] ] where @xmath39 , and the principal slow - roll parameter @xmath40 ( which is a measure of the acceleration of the universe ) , generally defined as @xmath41 is a constant during single fluid dominated era , related to the equation of state parameter , @xmath42 and we use it to characterize different cosmological eras . a schematic depiction of the expansion history of the universe assumed in this work , in terms of @xmath40 parameter and the conformal hubble rate is given in fig . [ expansion history ] and fig .
[ hubble rate ] , respectively .
parameter , consisting of three past cosmological eras inflation period , radiation dominated period , and matter dominated period and the today we are in a dark energy dominated period .
time @xmath43 corresponds to beginning of inflation , @xmath44 to the end of inflation , @xmath45 to the time of radiation - matter equality , and @xmath46 to the onset of dark energy domination period .
today we are at at @xmath47 .
the first two transitions are assumed to be fast in the sense that the scale of the duration of the transition @xmath48 satisfies @xmath49 .
we approximate the inflationary period by an exact de sitter one , @xmath50.,title="fig:",width=377 ] -0.5 cm throughout the expansion history of the universe .
hubble rates at the times of transitions satisfy the hierarchy @xmath51 . as in fig .
[ expansion history ] , time @xmath43 denotes the beginning of inflation , time @xmath52 the end of inflation , @xmath45 the end of radiation and beginning of matter dominated period ( radiation - matter equality ) , and time @xmath46 denotes the onset of dark energy domination.,title="fig:",width=377 ] -0.5 cm
this section introduces the model of a nonminimally coupled massive scalar field on cosmological space - time studied in this work .
the quantization of the scalar on flrw backgrounds is outlined , and the main quantities to be calculated expectation values of the energy - momentum tensor components are defined .
the choice of initial state is discussed .
the action for the massive nonminimally coupled scalar on curved space - time is @xmath53 \ , , \label{action : massive nonminimal scalar}\ ] ] where @xmath54 is the scalar field mass , and @xmath1 is the nonminimal coupling constant .
note that the sign convention we use implies @xmath55 \xrightarrow{d\rightarrow4}1/6 $ ] is the conformal coupling . on a flrw background
the lagrangian density takes the form @xmath56 \ , .\ ] ] in order to quantize this field we need to switch to the hamilton formalism .
first we define a canonically conjugate momentum , @xmath57 and then the hamiltonian via the legendre transform , @xmath58 \nonumber \\ = { } & \frac{a^{d-2}}{2 } \int\ !
d^{d-1}x \bigl [ a^{4 - 2d } \pi^2 + \bigl ( \vec{\nabla } \phi \bigr)^2 + ( ma)^2 \phi^2 + \xi a^2 r \ , \phi^2 \bigr ] \ , .\end{aligned}\ ] ] next we promote @xmath59 and @xmath60 to operators , and their poisson brackets to commutators , @xmath61 = i \delta^{d-1}(\vec{x } \!-\ !
\vec{x}^{\,\prime } ) \ , , \label{commphi1 } \\ & \bigl [ \hat{\phi}(\eta,\vec{x } ) , \hat{\phi}(\eta,\vec{x}^{\,\prime } ) \bigr ] = 0 = \bigl [ \hat{\pi}(\eta,\vec{x } ) , \hat{\pi}(\eta,\vec{x}^{\,\prime } ) \bigr ] \ , .
\label{commphi2}\end{aligned}\ ] ] the hamiltonian operator defined as @xmath62 now determines the dynamics via heisenberg equations for the field operators , @xmath63 = a^{2-d}(\eta ) \ , \hat{\pi}(\eta,\vec{x } ) \ , , \\ \hat{\pi}'(\eta,\vec{x } ) = { } & i \bigl [ \hat{h}(\eta ) , \hat{\phi}(\eta,\vec{x } ) \bigr ] = a^{d-2}(\eta ) \bigl [ \vec{\nabla}^2 - ( ma)^2 - \xi a^2(\eta ) r(\eta ) \bigr ] \hat{\phi}(\eta,\vec{x } ) \ , .\end{aligned}\ ] ] these two combined together yield the equation of motion for the field operator @xmath64 \biggr\ } \hat{\phi}(x ) = 0 \ , , \label{klein - gordon equation:1}\ ] ] where @xmath65 is the laplace operator .
this is just a klein - gordon equation , @xmath66 \hat{\phi } = 0 \ , , \label{klein - gordon equation:2}\ ] ] specialized to flrw , where @xmath67 $ ] is the dalembert operator .
this equation is standardly analyzed in fourier ( comoving momentum ) space , @xmath68 \ , , \label{fouriertransform}\ ] ] where @xmath69 and @xmath70 are the annihilation and creation operators , respectively , which satisfy the following commutation relations , @xmath71 = \delta^{d-1}(\vec{k } \!-\ !
\vec{k}^{\,\prime } ) \ , , \label{commb1 } \\ & \bigl [ \hat{b}(\vec{k } ) , \hat{b}(\vec{k}^{\,\prime } ) \bigr ] = 0 = \bigl [ \hat{b}^{\dag}(\vec{k } ) , \hat{b}^{\dag}(\vec{k}^{\,\prime } ) \bigr ] \ , , \label{commb2}\end{aligned}\ ] ] and @xmath72 is the mode function .
note the @xmath73 factor taken out in the definition of the fourier transform .
the commutation relations ( [ commphi1]-[commphi2 ] ) and ( [ commb1]-[commb2 ] ) require the wronskian of the mode function to be normalized as , @xmath74 the equation of motion satisfied by the mode function is the one for a harmonic oscillator with a time dependent frequency , @xmath75 u(k,\eta ) = 0 \ , , \label{modeeom}\ ] ] where @xmath76 \bigl [ 2\mathcal{h } ' + ( d\!-\!2 ) \mathcal{h}^2 \bigr ] \ , .
\label{m2}\ ] ] the state @xmath77 that we choose to examine we pick to be the one annihilated by the annihilation operator , @xmath78 , which implies there is no classical condensate ( hence the name spectator ) , @xmath79 this state respects the symmetries of the background space - time , namely homogeneity and isotropy , which is evident from requiring the mode function to depend only on the modulus of the comoving momentum . in order to completely specify this state one needs to specify the initial conditions for the mode function ( initial state ) , which we comment upon in subsection [ subsec : choice of state ] .
the energy - momentum tensor operator is defined as @xmath80}{\delta g^{\mu\nu}(x ) } \right|_{\phi\rightarrow\hat{\phi } } \nonumber \\
= { } & \partial_{\mu } \hat{\phi}(x ) \ , \partial_{\nu } \hat{\phi}(x ) - \frac{1}{2 } g_{\mu\nu}(x ) \bigl [ g^{\alpha\beta(x ) } \partial_{\alpha } \hat{\phi}(x ) \ , \partial_{\beta } \hat{\phi}(x ) \bigr ] - \frac{m^2}{2 } g_{\mu\nu}(x ) \hat{\phi}^2(x ) \nonumber \\ & + \xi \bigl [ g_{\mu\nu}(x ) - \nabla_{\mu } \nabla_{\nu } + g_{\mu\nu}(x ) \square \bigr ] \hat{\phi}^2(x ) \ , , \end{aligned}\ ] ] where @xmath81 is the einstein tensor , @xmath82 denotes the covariant derivative , and @xmath83 is the covariant dalembertian operator .
the expectation value of the energy momentum tensor operator with respect to the state defined in the previous section is diagonal , and is conveniently expressed in terms of energy density and pressure , @xmath84 \mathcal{h } ' |u|^2 \nonumber \\ & + 2m^2a^2 |u|^2 - \frac{1}{2 } \bigl [ d \!-\ !
4\xi(d \!-\ !
1 ) \bigr ] \mathcal{h } \frac{\partial}{\partial\eta } |u|^2 + \frac{1}{2 } \frac{\partial^2}{\partial\eta^2 } |u|^2 \biggr\ } \ , , \label{rhointegral}\end{aligned}\ ] ] @xmath85 \mathcal{h } ' |u|^2 \nonumber \\ & - \frac{1}{2 } \bigl [ d \!-\ ! 2 \!-\ !
4\xi(d \!-\ !
1 ) \bigr ] \mathcal{h } \frac{\partial}{\partial\eta } |u|^2 + \frac{(1 \!-\
! 4\xi)}{2 } \frac{\partial^2}{\partial\eta^2 } |u|^2 \biggr\ } \ , , \label{pintegral}\end{aligned}\ ] ] where we have used the equation of motion to write @xmath86 and eliminate @xmath87 in favor of @xmath88 and its derivatives .
we can take some derivatives out of the integral to write the and in a convenient way @xmath89 \bigl [ \mathcal{h } ' + \mathcal{h}\partial_\eta \bigr ] + \frac{1}{2 } \partial_\eta^2 \biggr ] \mathcal{i}_0 \biggr\ } \
, , \label{rhoini } \\ p_q = { } & \frac{a^{-d}}{(4\pi)^{\frac{d-1}{2 } } \ , \gamma\!\left ( \frac{d-1}{2 } \right ) } \biggl\ { \frac{2 \ , \mathcal{i}_1}{(d\!-\!1 ) } - \biggl [ \frac{1}{2}\bigl [ d \!-\ !
4\xi(d \!-\ !
1 ) \bigr ] \bigl [ \mathcal{h } ' + \mathcal{h}\partial_\eta \bigr ] + \frac{(1\!-\!4\xi)}{2 } \partial_\eta^2 \biggr ] \mathcal{i}_0 \biggr\ } \ , , \label{pini}\end{aligned}\ ] ] where we have defined the two integrals , @xmath90 finding a good approximation for these integrals is one of the two main technical tasks of this work .
understanding how the choice of the initial state affects our final results is important , and this is what we discuss next at some length .
we assume that the universe starts in a natural state defined on a global equal - time hyper - surface @xmath91 as the chernikov - tagirov - bunch - davies ( ctbd ) vacuum state in the ultraviolet ( uv ) .
this means that the mode function reduces to the flat space form in the deep uv , @xmath92 ( a more precise statement is given in appendix [ app : regularization and renormalization ] ) .
furthermore , we assume the state is suitably regulated in the infrared ( on super - hubble wavelengths ) .
that is namely necessary to regulate the infrared ( ir ) modes since attempting to impose the usual bunch - davies condition on the infrared modes would produce unphysical infrared divergences in the initial one - loop energy - momentum tensor .
infrared states can be regulated in various ways : ( i ) by choosing the global ctbd vacuum state associated with the epoch that precedes inflation in which the ctbd state is regular , ( ii ) by making the universe s ( initial ) equal - time hyper - surface compact , or ( iii ) by introducing a comoving ir cutoff .
the first prescription can be achieved by _
e.g. _ assuming a pre - inflationary radiation epoch @xcite , while the second on by imposing a positive constant spatial curvature ( @xmath93 ) on @xmath91 or by making @xmath91 compact by imposing periodic boundary conditions ( in the former case @xmath91 is a three - dimensional sphere @xmath94 while in the latter case @xmath91 is a 3-dimensional torus @xmath95 ) .
the third option is technically perhaps the simplest , and we employ it in this work .
it should be stressed that the point of view on this regularization is not to throw away the deep ir modes below the cutoff on principal grounds , but rather that they are smoothly suppressed under this scale and contribute negligibly to the observables .
then the leading approximation to this case is to introduce a sharp cutoff .
the deep ir suppression can be attributed to some physical process during or before inflation , or can be viewed as an approximation to the state obtained by placing the universe in a co - moving box .
one can show @xcite that , to leading order in powers of the ir comoving cutoff @xmath96 , ( expectation values of ) physical observables are correctly reproduced by the sudden cutoff approximation .
all these methods qualitatively agree . for ( i ) and
( ii ) this was shown in @xcite in the sense that qualitative dependence on the relevant physical scale is the same , where in the case of a pre - inflationary radiation epoch the relevant physical scale is the hubble parameter at the radiation - inflation transition , in the case when @xmath97 the relevant physical scale is @xmath98 and when @xmath99 the relevant physical scale is the comoving length of the torus @xmath100 .
note that the three aforementioned ways of regulating the infrared correspond to three ( very ) different physical situations . here
we use the simple cutoff regularization we _ effectively _ remove the modes below certain pivotal mode @xmath96 . in practice this is implemented by cutting of the integration of @xmath101 integrals at @xmath96 , @xmath102 in the limit of very small @xmath96 this can be shown to be equivalent ( up to corrections suppressed as @xmath103 ) to ( i ) mentioned above , with @xmath96 identified with @xmath104 , and is shown to be equivalent to ( ii ) here in sec . [
sec : energy density and pressure ] by comparison to @xcite .
the main point we are trying to make here is that , for a large class of initial states that are regular in the infrared one will get answers that qualitatively agree with the results obtained in this work , hence making the results of our analysis _ quite generic _ , _
i.e. _ to a large extent independent on the choice of the initial state .
the two main technical tasks of this work are ( i ) to solve for the time evolution of the mode function as it evolves through cosmological eras , and ( ii ) to perform the integrals and over these mode functions to obtain the backreaction energy density and pressure .
this section discusses these two issues from a more general point of view .
the mode function is organized in a convenient way .
relevant integration interval is identified , and a sudden transition approximation introduced for the contributing modes .
these considerations simplify following computations significantly .
when it comes to the evolution of the mode function , unfortunately , exact results are known only for a handful of flrw backgrounds .
there is a way to write down a general solution for an arbitrary flrw background , valid for all momentum scales @xcite , but it is difficult to make use of it practically .
luckily , for periods of constant @xmath40 ( out of which most of the history of expansion consists , fig .
[ expansion history ] ) the exact solutions are known in the massless limit , a convenient way to express them is in therms of chernikov - tagirov - bunch - davies ( ctbd ) mode function @xcite , @xmath105 where the index of the hankel function of the first kind @xmath106 is @xmath107 } \ , .\ ] ] these functions are defined to reduce to the positive - frequency form in the uv , and the ir is defined as an analytic continuation of the uv . the other linearly independent solution is a complex conjugate of . in the massive case ,
exact solutions are not known for constant @xmath40 periods , except in a few notable cases ( de sitter , and radiation dominated universe ) .
nevertheless , there is a way of making a controlled expansion of this function in the small ratio @xmath108 with which we will be concerned , which is sufficient for our purposes .
this expansion is presented in section [ sec : mode functions ] .
generally , the full mode function during a given constant @xmath40 period will be a linear combination of the ctbd mode functions , @xmath109 coefficients @xmath110 and @xmath111 are called the bogolyubov coefficients , and they have to satisfy @xmath112 as dictated by the wronskian normalization .
they are determined for each era by the initial conditions at the beginning of the given era , which are in turn given by the details of the transition from one era to another . if @xmath48 is a small time scale of the transition between periods , then for momenta above this scale the bogolyubov coefficients must reduce to @xmath113 nonadiabatically , meaning faster than any power of @xmath114 .
this is provided that the initial condition is of adiabatic order @xmath115 .
otherwise , if the initial state is of adiabatic order @xmath116 , it retains that property during the evolution @xcite .
this we can also infer from considerations of appendix [ app : regularization and renormalization ] .
the physical reason behind this conclusion is that the deep uv modes oscillate so fast so that their evolution is adiabatic . in the ir ,
bogolyubov coefficients are not universal as the deep uv are . on the contrary ,
they do depend on the details of the transition between the two periods of constant @xmath40 . in case of a fast transition
they are not so sensitive ( to leading order ) to the details of the transition , but rather depend just on the two periods connected by the transition .
this we show in the next subsection . in case of fast transitions between constant @xmath40 periods , @xmath117 ,
the evolution of the ir modes , @xmath118 , through the transition is well described by the so - called sudden transition approximation , where the @xmath40 parameter jumps discontinuously from one constant value to another .
physically , these modes are very slow compared to the transition scale , and the transition is effectively instantaneous for them .
more precisely , the transition is instantaneous for the ir modes to leading order in the expansion in the transition scale @xmath48 .
we stress that this is an approximation for the evolution of the ir modes , not a model for the background evolution , and should not be extrapolated to uv modes as a model for the background makes sense only if one takes the continuum limit of a series of such small transitions , as was done in @xcite . ] . here
we illustrate the sudden transition approximation on a specific example of transition between two periods @xmath119 and @xmath120 .
let the evolution of @xmath40 between two periods be given by @xmath121 where @xmath122 + \frac{\epsilon_1}{2 } \biggl [ 1 + \tanh\biggl ( - \frac{1}{1 + \frac{\eta - \eta_0}{\tau/2 } } + \frac{1}{1 - \frac{\eta - \eta_0}{\tau/2 } } \biggr ) \biggr ] \ , .
\label{epsilon transition}\ ] ] before the transition let the full mode function be @xmath123 with some known bogolyubov coefficients @xmath124 and @xmath125 . after the transition the mode function is @xmath126 the equation we are trying to solve is the harmonic oscillator one with a time dependent frequency , @xmath127 where @xmath128 and @xmath129 is defined in .
one can check the wkb applicability condition @xmath130 , and for which ranges of momenta is it satisfied , @xmath131 since @xmath132 by assumption , the only modes that evolve adiabatically are the ones for which at least @xmath133 . for the modes @xmath134 another approximation applies .
there @xmath135 is the largest scale in the hierarchy , and we can expand the evolution in powers of @xmath48 .
we do this by expanding @xmath136 function , @xmath137 now it is straightforward to match the two solutions and , which are just the continuity conditions for the mode function and its derivative , @xmath138 solving these conditions for bogolyubov coefficients yields @xmath139 + \mathcal{o}(\tau ) \
, , \label{suddenalpha } \\
\beta_1(k ) = { } & i \bigl [ u_0(k,\eta_0 ) u_1'(k,\eta_0 ) - u_0'(k,\eta_0 ) u_1(k,\eta_0 ) \bigr ] + \mathcal{o}(\tau ) \ , . \label{suddenbeta}\end{aligned}\ ]
] these two formulas comprise the sudden transition approximation for the bogolyubov coefficients and the evolution of the mode function .
in this section we analyze the integrals on general grounds
. they can not be evaluated exactly ( except for very simple mode functions @xcite ) , and we have to resort to approximation schemes .
we first organize the integrand in a way which separates the part containing all the uv divergences ( among other contributions ) , and the uv finite part containing possibly relevant ir contributions .
the contributions from different scales , _
i.e. _ different integration intervals are examined , and the relevant interval where the dominant contribution comes from is identified .
the analysis presented here greatly facilitates the evaluation of integrals , especially since we get away with not evaluating certain parts of integrals explicitly ( as was done in @xcite ) .
the integrands of integrals contain the mode function only as a modulus @xmath140 , which can be written out in terms of the ctbd mode functions of a given constant @xmath40 period , @xmath141 ^ 2 \ , , \label{u2organization}\ ] ] where was used . in the deep uv bogolyubov coefficients reduce to faster than power law ( at least exponentially ) .
therefore , the uv divergent structure of integrals is completely captured by the @xmath142 on the right in .
we split the integrals into two parts , @xmath143 where the ctbd part is @xmath144 which contains all the uv divergences , and the bogolyubov part , @xmath145 where the integrand is @xmath146^ * \ , .
\label{intbog}\ ] ] note that we take @xmath18 limit in the bogolyubov part , since it is manifestly uv finite because of the properties of the bogolyubov coefficients , which simplifies its evaluation .
we can immediately say a lot about the ctbd contribution . the way to compute it is to first split it in the uv and ir parts by introducing a uv cutoff @xmath147 , @xmath148 its uv part needs to be regularized and then renormalized as is outlined in appendix [ app : regularization and renormalization ] , and its contribution is given in ( [ rhoquv]-[pquv ] ) and . note that this uv contribution is dependent on the fiducial cutoff @xmath149 introduced by hand .
that dependence cancels exactly with the opposite one coming from the ir contribution ( which can be evaluated in @xmath18 from the start ) .
this computation is performed explicitly in subsection [ subsec : inflationary era ] for inflationary era .
the only dimensionful quantities the ctbd part can depend on are the evolving hubble rate and mass , and possibly on the ir regulator .
the dependence on the regulator actually must cancel out with same contribution from the bogolyubov part .
the total dependence on the regulator in the final answer can only come from the ctbd part and it must have the same structure as the initial state @xcite . therefore , in the small mass limit we can represent possible contributions to energy density and pressure as @xmath150 + \bigl [ \#_4 + \#_5 \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } + \#_6 \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4 } + \dots \bigr ] \biggr\ } \ , , \ ] ] where @xmath151 s stand for some numerical coefficients of order one .
this contribution is not relevant during the radiation or matter period .
its magnitude is tiny compared to the background . in matter period it also redshifts faster than the background . in radiation period
it does not , rather it redshifts at the same rate as the background fluid ( @xmath152 term is absent in this case ) , but its magnitude is tiny .
the contribution from this ctbd part is always negligible compared to the background , and therefore we can neglect it .
if there is a large effect , it must lie in the bogolyubov part .
that contribution we analyze generally in the next subsection .
we would like to argue on general grounds about the contributions from different momentum scales to this integral .
for the sake of simplicity , consider the transition between two periods of constant @xmath40 . before the transition , during period @xmath119 , the state was the ctbd one ( @xmath153 , @xmath154 ) .
the state after the transition , during period @xmath120 is dictated by the transition between periods , which is assumed to be fast ( @xmath155 ) .
we examine the two cases for the second period separately the decelerated case ( @xmath156 ) , and the accelerated case ( @xmath157 ) .
after the transition to the decelerating period the time evolving hubble rate @xmath158 drops below the one at the transition point @xmath159 ( see fig . [ hubble rate ] ) and eventually the hierarchy of scales depicted in figure [ hierarchy3 ] is reached .
this happens some time after the transition , which is the regime we are interested in .
we split the integration into three intervals , separated by @xmath160 and @xmath149 , according to this hierarchy .
the modes in the highest interval @xmath161 contribute negligibly since the bogolyubov coefficients are nonadiabatically suppressed there ( @xmath162 ) .
l5 cm -0.5
cm -0.5 cm the contribution of the middle interval can be estimated rather generally .
we start by noting that @xmath163 \bigl| u_1(k,\eta ) \bigr|^2 \ , .
\label{middleint}\end{aligned}\ ] ] the time dependent bd mode function may be expanded asymptotically as in since @xmath164 on this interval , @xmath165 + \mathcal{o}\bigl ( \frac{\mathcal{h}}{k } \bigr)^{\!4 } \biggr\}\ ] ] the momentum scales in question are much smaller than the scale of the transition @xmath135 , so the bogolyubov coefficients are well approximated by the sudden transition ones , and depend on three quantities : @xmath166 , @xmath159 , @xmath167 .
since we are interested in the small mass limit we may expand the bogolyubov coefficients in powers of @xmath168 , @xmath169 \ , , \ ] ] and analogously for @xmath170 , so to leading order they depend only on the ratio @xmath171 .
therefore , we can approximate with @xmath172 \ , .\ ] ] making a variable substitution @xmath173 puts this integral into a form which is suitable for further approximation , @xmath174 \ , .\ ] ] the integrand is now dimensionless , and the limits of integration are @xmath175 , and @xmath176 , so we may perform ( asymptotic ) expansions in these limits .
it might happen so that the leading order contribution is dominated by one of the cutoffs , but this contribution ( and in fact any other cutoff dependent one ) must cancel with the opposite contribution from another part of the full integral .
although tedious , this can be checked explicitly as was done in @xcite .
therefore , what we are interested in is the contribution independent off @xmath160 and @xmath149 , since that is the only one that remains after all the parts are added up . that contribution has the following form , @xmath177 this gives the contribution to the energy density ( and momentum ) of the form @xmath178 it is just a radiation - like contribution , and redshifts away faster than the background ( or at the same rate in the case of radiation era ) .
therefore we may safely neglect it as long as it is not too big before the start of radiation period .
the lower part of the integral has a chance to contribute something that does not redshift away faster than the background .
its exact contribution is not so straightforward to estimate , but if there is an interesting effect to be found it derives from this contribution .
therefore , it is the only one we need to examine , which we do in sec .
[ sec : energy density and pressure ] . for completeness , next we examine the accelerated case , @xmath179 . in the case of universe transitioning from one period of constant @xmath119 where the scalar was in a ctbd state , to an accelerating period of constant @xmath179 , where the transition was fast , we soon reach a hierarchy of scales shown in case a of fig . [ acceleratingscales ] , and afterwards the one in case b , as the conformal hubble rate continues to grow .
we treat these two cases separately .
-0.6 cm -0.4 cm + 5 mm case a + 3 mm the hierarchy of scales in this case is depicted on the left of fig .
[ acceleratingscales ] .
the contribution from the modes @xmath180 is negligible because of the nonadiabatic suppression of bogolyubov coefficients , just as in the decelerating case .
the contribution from the middle interval can be estimated as @xmath181 \bigl| u_1(k,\eta ) \bigr|^2 \label{midintac}\ ] ] in a similar way as in the decelerating case . here , because of the hierarchy of scales , we may expand the sudden transition bogolyubov coefficients for large momenta ( see @xcite for the expansion ) , @xmath182 the time dependent mode function depends on @xmath166 , @xmath158 and @xmath183 . since @xmath184 we can expand away its @xmath54-dependence so that the leading order term depends only on @xmath185 and @xmath158 . then , using the variable substitution @xmath186 in the middle integral [ midintac ] we can estimate it as @xmath187 \ , .\ ] ] as in the decelerating case , this integral can be expanded for small lower and large upper limit . neglecting as before terms dependent on the artificially introduced cutoffs ,
what can remain is a contribution @xmath188 .
now , these contribute to energy density and pressure as @xmath189 .
if the numerical coefficient is not exponentially large , this contribution is negligible compared to the background fluid energy density , and it only redshifts away faster than the background . only the lower integrals remains to be evaluated .
+ 5 mm case b + 3 mm in this case the reasoning of case a applies , we just need not examine the middle integral , since here it is shifted into the region where bogolyubov coefficients are nonadiabatically suppressed ( as depicted on the right of fig .
[ acceleratingscales ] ) , and hence contributes negligibly .
one needs to examine the same ( lower ) interval to look for the dominant contribution .
in this section we derive the ctbd mode functions for each of the cosmological periods .
exact solutions are known for the inflationary and radiation periods . while we can perform integrals over the exact inflationary mode functions , the radiation ones are too complicated , and need to be approximated .
an expansion in small @xmath190 to first subleading order suffices for our goals .
this expansion is performed in two ways .
firstly , the exact radiation period mode function is expanded , and the approximation valid for all momenta obtained .
secondly , the method for obtaining the approximation directly from the equation of motion ( without referring to the exact solution ) is introduced , and the approximation for the radiation period mode function derived .
this method is shown to reproduce the expansion of the exact result , which lends support for applying it in cases where the exact mode functions are not known .
the matter period mode functions are not known exactly , and we apply this method in order to find its expansion to first subleading order in small mass , valid for all momenta .
all the approximated mode functions derived in this section are simple enough so that integrals can be performed , and the energy density and pressure of quantum backreaction computed .
fortunately , the de sitter inflationary period ( @xmath50 ) ctbd mode functions are known exactly even in the massive case .
the equation of motion for the mode function is @xmath191u = 0 \ , .\ ] ] the positive frequency solution to this equation is @xmath192 where @xmath193 in this work we will be considering mass of the order of hubble rate today ( meaning that the ratio @xmath194 in inflation is extremely small ) , and nonminimal couplings @xmath195 .
more negative nonminimal couplings would lead to a too rapid growth of quantum fluctuations during inflation ( as will be shown by the end of this subsection ) , and a larger mass would mean that the field becomes very massive at some point during the cosmological evolution and starts contributing like dust to the expansion ( precluding it from having anything to do with de ) . for most of the range of these two parameters @xmath196 , which leads to an ir divergence .
this divergence has to be regulated somehow , since it signals that the state chosen is unphysical .
the practical method of regularization we choose is introducing a comoving ir cutoff @xmath96 .
the point of view we take on it is that it is a shortcut way of choosing a physical mode function , since the contribution of the modes below @xmath96 will be suppressed , and we neglect it from the start .
we will also need the small momentum ( @xmath197 ) expansion of the mode function in , @xmath198 \ , .
\label{ui_ir}\ ] ] in this subsection first the exact the radiation period ctbd mode function is derived , and then an expansion for @xmath199 performed ( valid for all momenta ) .
secondly , this small mass expansion is derived directly from the equation of motion without referring to the exact solution with the help of frobenius method .
this serves to introduce the method which can be applied to cases where an exact solution is not known .
the ctbd mode functions for radiation period are also known in the massive case , but are unfortunately too complicated for practical analytical computations .
the equation of motion for the modes is @xmath200 u = 0 \ , .
\label{eomrad1}\ ] ] in radiation period the scale factor and the conformal hubble rate are related as @xmath201 where the quantities with index 1 refer to the values at the beginning of radiation period , so the equation can be written as @xmath202 u = 0 \ , , \label{eomrad2}\ ] ] where we have defined @xmath203 by making a variable substitution @xmath204 the equation can be put in the form @xmath205 \biggr\ } \bigl ( s^{1/4 } u \bigr ) = 0 \ , , \ ] ] which is a whittaker equation @xcite , @xmath206 \biggr\ } \bigl ( s^{-1/4 } u \bigr ) = 0 \ , , \ ] ] with coefficients @xmath207 the properly normalized ( using the wronskian 13.14.30 from @xcite ) ctbd mode function is @xmath208 where @xmath209 is the whittaker function . by examining the uv expansion @xmath210 of ( corresponding to the large parameter expansion 9.229 from @xcite ) , @xmath211 \ , , \label{ctbdrad}\ ] ] we see that it indeed is the positive - frequency mode function ( the time - independent phase is irrelevant since it cancels out in all the physical quantities , the mode function is defined up to such a phase ) .
next we want to find the expansion of this function in small parameter @xmath212 , but valid for all momenta . in order to accomplish this
we view the function as a function of momenta .
in particular we represent it as a uniformly convergent power series in momenta .
the coefficients in this expansion are functions of @xmath54 and @xmath158 , and them we expand in this small ratio .
it is more convenient to express the whittaker function in terms of confluent hypergeometric functions , @xmath213 the radius of convergence of the power series representation of the confluent hypergeometric function is infinite so we may safely examine it , and do manipulations of it , @xmath214 where @xmath215 is the pochhammer symbol .
what we aim to do is to rewrite it as the power series in @xmath185 .
, and ultimately arrive at the same result .
expressing it as a function of @xmath185 is convenient though , since it is easy to take the massless limit for which mode functions are known and considerably simpler . ] in order to do this we write out the pochhammer symbol , @xmath216 where @xmath217-coefficients can be determined by writing out the pochhammer symbol . using this the power series representation of the confluent hypergeometric function can be rewritten as @xmath218 this power series is now straightforward to approximate this power series for small @xmath219 we simply throw away all the terms except the first three ones , @xmath220 this approximation can be seen to be valid for all the ranges of momenta . for @xmath221
this is just an expansion in @xmath183 which is the smallest scale in the hierarchy .
for @xmath222 the function is well described by a double expansion , in @xmath219 , and @xmath185 , so [ 1f1reorganized ] is good , it just retains more terms than necessary in this limit , which we neglect anyway .
now , we need only the first three @xmath217-coefficients introduced in , @xmath223 \ , .\end{aligned}\ ] ] the approximations for the confluent hypergeometric functions are then @xmath224 \frac{\mathcal{h}}{k } \sin\bigl ( \frac{k}{h } \bigr ) + \bigl [ \frac{\mathcal{h}^2}{6k^2 } - \frac{\mathcal{h}^4}{4k^4 } \bigr ] \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) \biggr\ } \ , , \\ & { } _ 1f_1 \bigl ( \frac{1}{4 } \!+\ ! \frac{ik^2}{4\overline{m}^2 } ; \frac{1}{2 } ; \frac{i\overline{m}^2}{\mathcal{h}^2 } \bigr ) \approx \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) + \bigl ( \frac{ma}{\mathcal{h } } \bigr ) \times \frac{i}{2 } \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) \nonumber \\ & \hspace{2 cm } + \bigl ( \frac{ma}{\mathcal{h } } \bigr)^2 \times \biggl\ { \bigl [ -\frac{1}{8 } \!+\ ! \frac{\mathcal{h}^2}{4k^2 }
\bigr ] \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) + \bigl [ \frac{1}{6 } - \frac{\mathcal{h}^2}{4k^2 } \bigr ] \frac{\mathcal{h}}{k } \sin \bigl ( \frac{k}{\mathcal{h } } \bigr ) \biggr\ } \ , .\end{aligned}\ ] ] after approximating the confluent hypergeometric functions in the small mass limit , we only need to approximate the exponential in , @xmath225 to arrive at an approximation for the full ctbd mode function in radiation period , @xmath226 \biggr\ } \nonumber \\ & + i^{1/4 } \sqrt{\frac{\pi}{2\overline{m } } } \ , \frac{e^{-\frac{\pi k^2}{8\overline{m}^2 } } } { \gamma\!\left ( \frac{3}{4 } \!+\ ! \frac{ik^2}{4\overline{m}^2 } \right ) } \biggl\ { \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) \nonumber \\ & \hspace{1 cm } + \bigl ( \frac{ma}{\mathcal{h } } \bigr)^2 \biggl [ - \frac{\mathcal{h}^2}{4k^2}\cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) - \bigl ( 1 \!-\ ! \frac{3\mathcal{h}^2}{2k^2 } \bigr ) \frac{\mathcal{h}}{6k } \sin\bigl ( \frac{k}{\mathcal{h } } \bigr ) \biggr ] \biggr\ } \ , .
\label{ctbdradapprox}\end{aligned}\ ] ] numerical comparisons with the exact ctbd mode function show this is a very good approximation for small mass , @xmath227 , for all the ranges of momenta .
note that we can not expand the time - independent coefficients multiplying the curly brackets if we want this approximation to be valid for both small and large momenta .
one important property that these coefficients satisfy is @xmath228^ * \times \biggl [ i^{1/4 } \sqrt{\frac{\pi}{2\overline{m } } } \ , \frac{e^{-\frac{\pi k^2}{8\overline{m}^2 } } } { \gamma\!\left ( \frac{3}{4 } \!+\ ! \frac{ik^2}{4\overline{m}^2 } \right ) } \biggr ] \biggr\ } = \frac{1}{2 } \ , . \label{coeffrelation}\ ] ] in the next subsection we develop a method to obtain this approximation directly from the equation of motion
. we will be able to determine the time dependent functions in the curly brackets in , but not the time - independent coefficients in front of the brackets .
however , the important property will follow . here
we wish to derive the approximation directly from the equation of motion .
the reason for doing this alongside having an exact solution is to establish the approximation method on an example where we can compare and test it .
then afterwards we will apply this method to the matter period case where exact solution is not available .
the method is in the spirit of the way in which we derived the small mass expansion of the mode function in the previous subsection .
we use the frobenius method @xcite to find the power series solution to the equation of motion , and then reorganize it to write it as a double power series in @xmath219 and @xmath185 , which is then straightforward to approximate . starting from the equation of motion , and making a variable substitution @xmath229
puts the equation in the form @xmath230 u = 0 \ , .
\label{radeom3}\ ] ] the frobenius method consists of assuming a power series solution , @xmath231 plugging it in the equation , and solving order by order for the coefficients . the resulting equation , ordered in powers of @xmath232 is @xmath233 z^{\lambda } \nonumber \\ & + \bigl [ ( 3 \!+\ !
\lambda ) ( 2 \!+\ !
\lambda ) c^{(\lambda)}_3
+ c^{(\lambda)}_1 \bigr ] z^{\lambda+1 } + \sum_{n=0}^{\infty } \biggl [ ( n \!+\ ! \lambda \!+\ !
4 ) ( n \!+\ ! \lambda \!+\ !
3 ) c^{(\lambda)}_{n+4 } + c^{(\lambda)}_{n+2 } + \frac{\overline{m}^4}{k^4 } c^{(\lambda)}_n \biggr ] z^{\lambda+2+n } \ , .\end{aligned}\ ] ] coefficients multiplying different powers of @xmath232 must vanish independently , which gives us an infinite set of equations .
the leading order gives the so - called indical polynomial @xmath234 whose roots @xmath235 distinguish between the two linearly independent solutions .
the leading order coefficient @xmath236 is the overall normalization of the function , and can not be determined by this method ( stemming from the fact that equation is linear and homogeneous ) .
the second order requires @xmath237 which is satisfied by setting @xmath238 .
coefficient @xmath239 is undetermined from this equation , and can be chosen arbitrarily , so we set it to zero for convenience .
in fact , picking a nonzero value of it corresponds to choosing a different linear combination of independent solutions . ]
it is also straightforward to see that all the rest of the odd coefficients must vanish as well , @xmath240 this leaves the even coefficients to be determined .
order @xmath241 gives @xmath242 and the remaining coefficients are determined by the recurrence relation , @xmath243 we do not bother to solve this recurrence relation exactly , since the order of approximation we are after does not require it .
instead , we note that the coefficients have the following form , @xmath244 plugging in this into the initial power series , and reorganizing , gives the desired double power series ( remember that @xmath245 ) , @xmath246 the approximation to first subleading order in small mass now consists of keeping just @xmath247 and @xmath248 terms , @xmath249 what remains is to solve for the needed @xmath250-coefficients by using and the recurrence relation , @xmath251 \ , , \end{aligned}\ ] ] and to resum the power series in using these coefficients .
the two linearly independent solutions ( @xmath252 ) that we find are @xmath253 \frac{\mathcal{h}^2}{4k^3 } \sin\bigl ( \frac{k}{\mathcal{h } } \bigr ) + \bigl [ 1 \!-\ ! \frac{3\mathcal{h}^2}{2k^2 } \bigr ] \frac{\mathcal{h}}{6k^2}\cos\bigl ( \frac{k}{\mathcal{h } } \bigr)\biggr\ } \ , , \label{vr1 } \\
v_{r2 } = { } & \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) + \bigl ( \frac{ma}{\mathcal{h } } \bigr)^2 \times \biggl\ { -\frac{\mathcal{h}^2}{4k^2 } \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) - \bigl [ 1 \!-\ ! \frac{3\mathcal{h}^2}{2k^2 } \bigr ] \frac{\mathcal{h}}{6k } \sin\bigl ( \frac{k}{\mathcal{h } } \bigr ) \biggr\ } \ , , \label{vr2}\end{aligned}\ ] ] where we have picked the normalizations @xmath254 and @xmath255 for convenience , and so that the @xmath256 limit is well defined for both functions .
these two functions are exactly the ones in curly brackets in that were found by expanding the exact solution in small mass .
the ctbd mode function in radiation period is some linear combination of and , @xmath257 since ctbd mode functions are assumed to satisfy the wronskian normalization , it is easy to compute that the coefficients above must satisfy @xmath258 = \frac{1}{2 } \ , .
\label{radcoeffproperty}\ ] ] this is in fact an exact relation between these coefficients , valid to all orders in @xmath54 , which we have already calculated from the exact solution in .
we can not say more about these coefficients just based on the equation of motion , but luckily we do not have to for the purposes of computing the backreaction energy - momentum tensor , will be the only property needed .
later we will need an ir expansion of the mode functions and which we include here , @xmath259
+ \mathcal{o}\bigl ( \frac{k}{\mathcal{h } } \bigr)^{\!2 } \ , , \label{vr1_ir } \\
v_{r2 } \approx { } & \biggl [ 1 - \frac{1}{12 } \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4 } \biggr ] + \mathcal{o}\bigl ( \frac{k}{\mathcal{h } } \bigr)^{\!2 } \ , .
\label{vr2_ir}\end{aligned}\ ] ] here we apply the method introduced in the previous subsection from the start since the exact solution for the mode function is not known . during matter period ( @xmath260 ) the background satisfies @xmath261 .
the equation of motion for the modes , when a variable substitution @xmath262 is put into the form @xmath263 u = 0 \ , , \label{matteom}\ ] ] where we have defined a mass parameter @xmath264^{1/3 } \ , , \ ] ] which satisfies @xmath265 we do not know the exact solutions of the equation of motion .
that is why we will resort to the approximation scheme for the small mass expansion developed in the previous subsection .
as before , we use the frobenius method to obtain a power series solution to the equation , @xmath266 organizing the equation in powers of @xmath232 yields @xmath267 c_0^{(\lambda ) } z^{\lambda-2 } + \bigl [ ( \lambda \!+\ !
1 ) \lambda \!-\ ! 2(1 \!-\ !
6\xi ) \bigr ] c_1^{(\lambda ) } z^{\lambda-1 } \nonumber \\ & + \biggl\ { \bigl [ ( \lambda \!+\ ! 2)(\lambda \!+\ !
1 ) \!-\ ! 2(1 \!-\ !
6\xi ) \bigr ] c_2^{(\lambda ) } + c_0^{(\lambda ) } \biggr\ } z^{\lambda } \nonumber \\ & + \biggl\ { \bigl [ ( \lambda \!+\ ! 3)(\lambda \!+\ !
2 ) \!-\ ! 2(1 \!-\ !
6\xi ) \bigr ] c_3^{(\lambda ) } + c_1^{(\lambda ) } \biggr\ } z^{\lambda+1 } \nonumber \\ & + \biggl\ { \bigl [ ( \lambda \!+\ ! 4)(\lambda \!+\ !
3 ) \!-\ ! 2(1 \!-\ !
6\xi ) \bigr ] c_4^{(\lambda ) } + c_2^{(\lambda ) } \biggr\ } z^{\lambda+2 } \nonumber \\ & + \biggl\ { \bigl [ ( \lambda \!+\ ! 5)(\lambda \!+\ !
4 ) \!-\ ! 2(1 \!-\ !
6\xi ) \bigr ] c_5^{(\lambda ) } + c_3^{(\lambda ) } \biggr\ } z^{\lambda+3 } \nonumber \\ & + \sum_{n=0}^{\infty }
\biggl\ { \bigl [ ( \lambda \!+\ ! 6 \!+\ ! n)(\lambda \!+\ ! 5 \!+\ !
n ) \!-\ ! 2 ( 1 \!-\ !
6\xi ) \bigr ] c_{n+6}^{(\lambda ) } + c_{n+4}^{(\lambda ) } + \frac{\widetilde{m}^6}{16k^6 } c_n^{(\lambda ) } \biggr\ } \ , .\end{aligned}\ ] ] coefficients multiplying different powers of @xmath232 must vanish independently .
the order @xmath268 gives the indicial polynomial , @xmath269 whose solutions are @xmath270 \equiv \frac{1}{2 } + \nu \ , , \label{lambda1 } \\
\lambda_2 = \frac{1}{2 } \bigl [ 1 - \sqrt{1 \!+\ ! 8(1 \!-\ !
6\xi ) } \bigr ] \equiv \frac{1}{2 } - \nu \ , , \label{lambda2}\end{aligned}\ ] ] and @xmath271 is the overall normalization constant .
note that in and above we have introduced the definition of @xmath272 , @xmath273 which would be the index of the hankel functions in ctbd mode function of the matter period in the massless limit .
the next order requires that @xmath274 , and in fact all the odd coefficients must vanish , @xmath275 orders @xmath241 and @xmath276 give @xmath277 which serve as initial conditions for the recurrence relation @xmath278 the indicial polynomial was used to simplify the denominators of the above expressions for the coefficients . in a similar fashion as for the case of radiation period in the previous section
, the coefficients in the expansion can be seen from to have the form @xmath279 where we will not need to solve for all the @xmath250-coefficients . plugging these into the power series , and reorganizing gives @xmath280 \bigl ( \frac{2k}{\mathcal{h } } \bigr)^{\!6(n - s ) } \ , , \ ] ] which is straightforward to approximate in the @xmath227 limit , by keeping only @xmath281 and @xmath282 terms , @xmath283 now the @xmath250-coefficients we need can be found from , and they are @xmath284 \ , .\end{aligned}\ ] ] resumming the series yields the two linearly independent solutions to first subleading order in small @xmath219 , @xmath285 j_{1+\nu } \bigl ( \frac{2k}{\mathcal{h } } \bigr ) \nonumber \\ & \hspace{2.5 cm } + ( 2\!-\!\nu ) \biggl [ 3 \!-\ ! ( 1\!-\!\nu)(2\!+\!\nu ) \frac{\mathcal{h}^2}{k^2 } \biggr ] \frac{\mathcal{h}}{k } j_{2+\nu } \bigl ( \frac{2k}{\mathcal{h } } \bigr ) \biggr\ } + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4 } \ , , \label{vm1 } \\
v_{m2 } = { } & \gamma(1\!-\!\nu ) k^{\nu } \mathcal{h}^{-1/2 } \ , j_{-\nu } \bigl ( \frac{2k}{\mathcal{h } } \bigr ) \nonumber \\ & + \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } \times \frac{\gamma(1\!-\!\nu)}{30 } k^{-1+\nu } \mathcal{h}^{1/2 } \biggl\ { \biggl [ -6 \!+\ ! ( 1\!+\!\nu)(2\!+\!\nu ) \frac{\mathcal{h}^2}{k^2 } \biggr ] j_{1-\nu}\bigl ( \frac{2k}{\mathcal{h } } \bigr ) \nonumber \\ & \hspace{2.5 cm } + ( 2\!+\!\nu ) \biggl [ 3 \!-\ ! ( 1\!+\!\nu)(2\!-\!\nu ) \frac{\mathcal{h}^2}{k^2 } \biggr ] \frac{\mathcal{h}}{k } j_{2-\nu } \bigl ( \frac{2k}{\mathcal{h } } \bigr ) \biggr\ } + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4 } \ , , \label{vm2}\end{aligned}\ ] ] where a convenient overall normalization was chosen , @xmath286 and @xmath287 , and @xmath272 was defined in .
the ctbd mode function in matter period is some linear combination of the two independent solutions above , @xmath288 we can not determine them just from the equation of motion , but they satisfy @xmath289 = \frac{1}{2\nu } \ , , \label{mattcoeffproperty}\ ] ] which follows from the wronskian normalization , and the wronskian of the functions and .
we expect this to be an exact relation in the same way as the analogous one for radiation period , but we have not checked it explicitly .
we will not need any other properties of the coefficients other than in order to compute the energy - momentum tensor of the backreaction in matter period .
later we will need also an ir expansion of the mode functions and , which we give here , @xmath290
+ \mathcal{o}\bigl ( \frac{k}{\mathcal{h } } \bigr)^{\!2 } \biggr\ } \ , , \\ v_{m2 } \approx { } & \mathcal{h}^{-1/2+\nu } \biggl\ { \biggl [ 1 - \frac{1}{3(3\!-\!\nu ) } \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4 } \biggr ] + \mathcal{o}\bigl ( \frac{k}{\mathcal{h } } \bigr)^{\!2 } \biggr\ } \ , .\end{aligned}\ ] ]
this section is devoted to computing the leading contributions to integrals , using the rationale from section [ sec : energy density and pressure integrals ] , and the mode functions derived in section [ sec : mode functions ] .
the computation is made for all three cosmological eras ( fig .
[ expansion history ] ) , up until the onset of de domination .
the final answers are leading contributions in the small ratios of physical parameters ( which satisfy a hierarchy from fig .
[ hubble rate ] ) .
also , the computations are restricted to the regions long enough after the transition periods so that this hierarchy can be exploited . at the end of each subsection
the minimally coupled limit is discussed , and compared to @xcite , as well as the limits of small nonminimal coupling which is the main focus of this work . for exact de sitter inflationary era we can actually evaluate the integrals for the energy - momentum tensor exactly , and
there is no need to resorting to approximations .
first we compute the ir part of , using the ctbd mode function , @xmath291 \nonumber \\ & + \frac{2^{2\nu_i-3 } \ , \gamma^2(\nu_i ) } { \pi \left ( \nu_i \!-\ ! \frac{3}{2 } \right ) } \mathcal{h}^{2\nu_i-1 } k_0^{3 - 2\nu_i } \bigl [ 1 + \mathcal{o}\bigl ( \frac{k_0}{\mathcal{h } } \bigr)^{\!2 } \bigr ] \ , , \\ \mathcal{i}_1^{ir } = { } & \int\limits_{k_0}^{\mu } \ !
k^4 \bigl| u_i(k,\eta ) \bigr|^2 \nonumber \\ = { } & \frac{\mu^4}{8 } + \frac{\mathcal{h}^2\mu^2}{8 } \bigl ( \nu_i^2 \!-\ ! \tfrac{1}{4 } \bigr ) + \frac{3\mathcal{h}^2}{16 } \bigl ( \nu_i^2 \!-\ ! \tfrac{1}{4 } \bigr ) \bigl ( \nu_i^2 \!-\ ! \tfrac{9}{4 } \bigr ) \ln\bigl ( \frac{\mu}{\mathcal{h } } \bigr ) \nonumber \\ & + \frac{\mathcal{h}^4}{64 } \bigl ( \nu_i^2 \!-\ ! \tfrac{1}{4 } \bigr ) \bigl ( \nu_i^2 \!-\ ! \tfrac{9}{4 } \bigr ) \biggl [ 12\ln2 \!-\ ! 7 \!-\
! 6\psi\bigl ( - \tfrac{3}{2}\!-\!\nu_i \bigr ) \!-\ ! 6\psi\bigl ( -\tfrac{3}{2}\!+\!\nu_i \bigr ) \biggr ] \nonumber \\ & + \frac{2^{2\nu_i-3 } \ , \gamma^2(\nu_i ) } { \pi \left ( \nu_i \!-\ ! \frac{5}{2 } \right ) } \mathcal{h}^{2\nu_i-1 } k_0^{5 - 2\nu_i } \bigl [ 1 + \mathcal{o}\bigl ( \frac{k_0}{\mathcal{h } } \bigr)^{\!2 } \bigr ] \ , .\end{aligned}\ ] ] plugging these integrals into expressions and gives us the ir contributions to energy density and pressure during inflationary period .
combining these with the uv contributions and specialized to @xmath292 ( including the conformal anomaly as well ) , the dependence on the artificially introduced uv cutoff cancels as promised , and the physical renormalized quantity remains , @xmath293 \bigl ( \frac{m}{h } \bigr)^{\!2 } - \bigl [ \frac{1\!+\!2c_0}{4 } \bigr ] \bigl ( \frac{m}{h_i } \bigr)^{\!4 } \biggr\ } \nonumber \\
& + \frac{4^{\nu_i-2}\ , \gamma^2(\nu_i ) } { \pi^3 \left ( \nu_i\!-\!\frac{3}{2 } \right ) } \biggl [ \nu_i \bigl ( \nu_i \!-\ ! \tfrac{3}{2 } \!+\ ! 6\xi \bigr ) + \bigl ( \frac{m}{h_i } \bigr)^{\!2 } \biggr ] h_i^4 \bigl ( \frac{a_0h_i}{k_0 } \bigr)^{2\nu_i-3 } \ , , \label{rhoinf } \\
p_q = { } & \frac{h_i^4}{32\pi^2 } \biggl\ { ( 1 \!-\ ! 6\xi)^2 - \frac{1}{30 } + 2(1 \!-\ !
6\xi ) \bigl [ \frac{1\!-\!c_0}{2 } \bigr ] \bigl ( \frac{m}{h_i } \bigr)^{\!2 } + \bigl [ \frac{1\!+\!2c_0}{4 } \bigr ] \bigl ( \frac{m}{h_i } \bigr)^{\!4 } \biggr\ } \nonumber \\ & + \frac{4^{\nu_i-2}\ , \gamma^2(\nu_i ) } { \pi^3 \left ( \nu_i\!-\!\frac{3}{2 } \right ) } \nu_i \bigl [ \nu_i \!-\ ! \tfrac{3}{2 } \!-\ ! 4\xi ( \nu_i \!-\ !
2 ) \bigr ] h_i^4 \bigl ( \frac{a_0h_i}{k_0 } \bigr)^{2\nu_i-3 } \ , , \label{pinf}\end{aligned}\ ] ] where @xmath294 and @xmath295 is given in . in the expressions
above we have reverted to using the physical hubble rate @xmath296 , assumed to be constant during inflation ( @xmath292 ) , in order to make the expression more transparent .
the ir cutoff @xmath96 in and ( and in the results to follow ) can be given physical meaning by relating it to the hubble scale at the beginning of inflation @xmath159 . in @xcite ( and @xcite ) , the ir regularization method employed was matching the inflationary period onto a pre - inflationary radiation - dominated period , where ir issues are absent ( due to vanishing ricci scalar ) , so the hubble rate at the beginning of inflation was explicitly introduced . by specializing that result to @xmath292 , and comparing it to the one above , we see that @xmath297 of course . in this paper we will be dealing with
the nonminimal coupling restricted to @xmath298 , in which case on de sitter the two scales coincide to leading order in @xmath1 , @xmath299 .
therefore , from now on we will be making this identification .
setting the nonminimal coupling @xmath1 to zero , and working in the small mass limit gives @xmath300 \ , , \\
p_q = { } & \frac{119h_i^4}{960\pi^2 } - \frac{3 h_i^4}{16\pi^2 } \biggl [ 1 - e^{-\frac{2}{3}\frac{m^2}{h_i^2}n } \biggr ] \ , , \end{aligned}\ ] ] for the backreaction energy density and pressure , which is a standard result . in the expressions
above @xmath301 stands for the number of e - foldings from the beginning of inflation , @xmath302 .
the first term corresponds to the energy density and pressure of an exactly massless scalar in a ctbd state during de sitter inflation @xcite .
the second term is a contribution from @xmath303 , whose behavior is well known for ( slow roll ) inflationary backgrounds @xcite . for an extremely long inflation
the energy density and pressure saturate to @xmath304 and contribute just a tiny correction to the effective cosmological constant ( determined by the expansion rate ) .
this limit was used in @xcite in the context of late - time quantum backreaction . for a `` short '' inflation the backreaction at the end of inflation can be approximated to be @xmath305 where here and henceforth @xmath306 is the total number of e - foldings of inflation .
but @xmath306 does not have to be small , in fact it can still be very large if @xmath54 is very small compared to @xmath307 .
this limit proved better when constructing a de model based on backreaction @xcite .
this is effectively a massless limit of the full result and , and coincides with the results in @xcite , @xmath309 \ , , \qquad p_q \approx - \rho_q \ , .
\label{backinf}\ ] ] in the end we will be interested in this range of parameters during inflation , when we try to construct a model in which the quantum backreaction is small throughout the expansion history , and becomes large only at the onset of the de - dominated period ( fig .
[ expansion history ] ) .
we see that the backreaction is negative and grows in amplitude exponentially with @xmath301 during inflation , and how much it grows depends on the value of nonminimal coupling and the duration of inflation .
since we want the backreaction to remain perturbative during inflation , this imposes a constraint on the @xmath310 parameter space depicted in fig .
[ inflconstraint ] , which was derived by requiring @xmath311 . , and the total number of e - foldings of inflation @xmath306 .
the bold curve corresponds to the condition @xmath312 at the end of inflation ( @xmath313 ) , and the shaded region to situations when quantum backreaction starts to dominate even before the end of inflation .
this part of parameter space is not examined in this work , but rather concentrate on the white region where quantum backreaction stays perturbative during inflation .
the dashed line represents the requirement on the minimal duration of inflation .
it follows that we restrict ourselves to considering @xmath314 in this work.,title="fig:",width=377 ] -0.5 cm this implies for the number of e - foldings , @xmath315 \ , , \label{boundinf}\ ] ] where the dimensionful units were restored , and @xmath316 is the planck energy , and the inflationary hubble scale is taken to be @xmath317 . although strong backreaction in inflation would be very interesting to study in its own right ( especially since its energy density has a negative sign which would work towards slowing down inflation ) , here we restrict ourself to studying just the de scenarios , for which we assume small backreaction at the end of inflation .
some time after the transition to the radiation period the hierarchy of scales @xmath318 is reached ( together with the assumed @xmath319 ) .
the relevant contribution to integrals is @xmath320 as was established in subsection [ subsubsec : decelerating period ] , with @xmath321 and the integrand , as defined in , is @xmath322^ * \ , .
\label{radintegrand}\ ] ] the bogolyubov coefficients in this integrand are determined by the fast transition from the inflationary period to the radiation one . for the scales integrated over in they
are well approximated by the sudden transition ones and , @xmath323 \ , , \\
\beta_r(k ) = { } & i \bigl [ u_i(k,\eta_1 ) u_r'(k,\eta_1 ) - u_i'(k,\eta_1 ) u_r(k,\eta_1 ) \bigr ] \ , .\end{aligned}\ ] ] the inflationary ctbd mode function in bogolyubov coefficients above is given by , and the radiation ctbd mode function by .
for the range of integration in , the mode functions inside of bogolyubov coefficients are in fact very well described by the small momentum limit ( on top of small mass limit ) .
we use this to simplify the integrand before integration . using the ir expansion for the inflationary ctbd mode function
, it follows that @xmath324 and the bogolyubov coefficients simplify to @xmath325 \equiv - \bigl| u_i(k,\eta_1 ) \bigr| \widetilde{\beta}_r^*(k ) \
\beta_r(k ) \approx { } & \bigl| u_i(k,\eta_1 ) \bigr|
\bigl [ u_r'(k,\eta_1 ) - \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) \mathcal{h}_1 u_r(k,\eta_1 ) \bigr ] \equiv \bigl| u_i(k,\eta_1 ) \bigr| \widetilde{\beta}_r(k ) \ , , \end{aligned}\ ] ] and to leading order satisfy @xmath326 which implies the following simplification for the integrand , @xmath327 \biggr\}^2 \ , .\ ] ] this integrand is further simplified by considering the small mass limit @xmath328 , using the approximate mode functions and , @xmath329 \approx { } & \im \bigl [ a_r^*(k , m ) b_r(k , m ) \bigr ] \times \biggl\ { v_{r2}'(k,\eta_1 ) v_{r1}(k,\eta ) - v_{r1}'(k,\eta_1 ) v_{r2}(k,\eta ) \nonumber \\ & - \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) \mathcal{h}_1 \bigl [ v_{r2}(k,\eta_1 ) v_{r1}(k,\eta ) - v_{r1}(k,\eta_1 ) v_{r2}(k,\eta ) \bigr ] \biggr\ } \nonumber \\
= { } & \frac{1}{2 } \biggl\ { v_{r2}'(k,\eta_1 ) v_{r1}(k,\eta ) - v_{r1}'(k,\eta_1 ) v_{r2}(k,\eta ) \nonumber \\ & - \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) \mathcal{h}_1 \bigl [ v_{r2}(k,\eta_1 ) v_{r1}(k,\eta ) - v_{r1}(k,\eta_1 ) v_{r2}(k,\eta ) \bigr ] \biggr\ } \
, , \end{aligned}\ ] ] where the property was used , and it is the only place where we need to refer to coefficients @xmath330 and @xmath331 , no matter how complicated they may be . furthermore , we may expand this expression to leading order in @xmath332 because of hierarchy , @xmath329 \approx { } & - \frac{1}{2 } \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) \biggl\ { \frac{\mathcal{h}}{k } \sin\bigl ( \frac{k}{\mathcal{h } } \bigr ) + \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } \bigl ( \frac{\mathcal{h}}{k } \bigr)^{\!2 } \biggl [ - \bigl ( 1 \!-\ ! \frac{\mathcal{h}^2}{k^2 } \bigr ) \frac{\mathcal{h}}{4k } \sin\bigl ( \frac{k}{\mathcal{h } } \bigr ) \nonumber \\ & + \frac{1}{6 } \bigl ( 1 \!-\ ! \frac{3\mathcal{h}^2}{2k^2 } \bigr ) \cos\bigl ( \frac{k}{\mathcal{h } } \bigr ) \biggr ] + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4 } \biggr\ } \bigl ( \frac{\mathcal{h}_1}{\mathcal{h } } \bigr ) \bigl [ 1 + \mathcal{o}\bigl ( \frac{\mathcal{h}}{\mathcal{h}_1 } \bigr ) \bigr ] \ , , \end{aligned}\ ] ] so that now , after plugging in , the full integrand is well approximated by @xmath333 \biggr\ } \ , .
\label{radintegrandapprox}\end{aligned}\ ] ] finally , we can perform the integrals using this approximated integrand .
the result we expand according to the hierarchy , @xmath334 \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{\!3 - 2\nu_i } \bigl [ 1 + \mathcal{o } \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{\!2 } \bigr ] \nonumber \\ & - \frac{\gamma(1\!-\!2\nu_i)}{2^{1 - 2\nu_i } } \sin(\pi\nu_i ) \biggl [ 1 - \frac{8 \gamma(-3\!-\!2\nu_i)}{3\gamma(1\!-\!2\nu_i ) } \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) ( 2\nu_i^2 \!+\ ! 4\nu_i \!+\ !
3 ) \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4 } \biggr ] \nonumber \\ & - \frac{1}{2\left ( \nu_i \!-\ ! \frac{1}{2 } \right ) } \bigl ( \frac{\mathcal{h}}{\mu } \bigr)^{2\nu_i-1 } \bigl [ 1 + \mathcal{o } \bigl ( \frac{\mathcal{h}}{\mu } \bigr)^{\!2 } \bigr ] \biggr\ } \mathcal{h}^2 \bigl ( \frac{\mathcal{h}_1}{\mathcal{h } } \bigr)^{\!2\nu_i+1 } \bigl [ 1 + \mathcal{o } \bigl ( \frac{\mathcal{h}}{\mathcal{h}_1 } \bigr ) \bigr ] \ , , \end{aligned}\ ] ] @xmath335 \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{\!5 - 2\nu_i } \bigl [ 1 + \mathcal{o } \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{\!2 } \bigr ] \nonumber \\ & + \frac{\gamma(3\!-\!2\nu_i)}{2^{3 - 2\nu_i } } \sin(\pi\nu_i ) \biggl [ 1 - \frac{8 \gamma(-1\!-\!2\nu_i)}{3\gamma(3\!-\!2\nu_i ) } \bigl ( \nu_i \!-\ ! \tfrac{3}{2 } \bigr ) ( 2\nu_i^2 \!+\ !
1 ) \bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!4}\biggr ] \nonumber \\ & - \frac{1}{2\left ( \nu_i \!-\ ! \frac{3}{2 } \right ) } \bigl ( \frac{\mathcal{h}}{\mu } \bigr)^{2\nu_i-3 } \bigl [ 1 + \mathcal{o } \bigl ( \frac{\mathcal{h}}{\mu } \bigr)^{\!2 } \bigr ] \biggr\ } \mathcal{h}^4 \bigl ( \frac{\mathcal{h}_1}{\mathcal{h } } \bigr)^{\!2\nu_i+1 } \bigl [ 1 + \mathcal{o } \bigl ( \frac{\mathcal{h}}{\mathcal{h}_1 } \bigr ) \bigr ] \ , .\end{aligned}\ ] ] then plugging them into and gives the backreaction energy density and pressure ( in the small mass limit ) , @xmath336 \nonumber \\ & \hspace{0.5 cm } - \frac{\left ( \nu_i \!-\ ! \frac{1}{2 } \right ) \gamma(\nu_i ) } { \left ( \nu_i \!-\ ! \frac{3}{2 } \right ) } \bigl ( \frac{\mathcal{h}_0}{\mu } \bigr)^{\ ! 2\nu_i - 3 } \bigl ( \frac{a_1}{a } \bigr)^{\!4 } - \sqrt{\pi } ( \nu_i \!-\ !
1 ) \gamma\bigl ( \tfrac{3}{2 } \!-\ !
\nu_i \bigr ) \biggl [ ( 1 \!-\ ! 6\xi)\bigl ( \frac{a_1}{a } \bigr)^{\ ! 4 } \nonumber \\ & \hspace{0.5 cm } - \frac { ( 1 \!-\ !
6\xi ) \left ( \nu_i \!-\ ! \frac{3}{2 } \right ) + 3\nu_i \left ( \nu_i \!+\ ! \frac{3}{2 }
\right ) + 2(1\!-\!6\xi)\nu_i^2 \left ( \nu_i \!+\ ! \frac{3}{2 } \right ) } { 6\nu_i(\nu_i \!-\!1)\left ( \nu_i \!+\ ! \frac{1}{2 } \right ) \left ( \nu_i \!+\ ! \frac{3}{2 } \right ) } \bigl ( \frac{m}{h_i } \bigr)^{\!2 } \biggr ] \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{\ ! 3 \!-\ !
2\nu_i } \biggr\ } \ , , \label{rhoqm } \\
p_q \approx { } & \frac{\gamma(\nu_i)h_i^4}{2^{5 - 2\nu_i}\pi^3 } \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) e^{(2\nu_i \!-\ ! 3)n_i } \biggl\ { \frac{\left ( \nu_i \!-\ ! \frac{1}{2 } \right ) \gamma(\nu_i ) } { \left ( \nu_i \!-\ ! \frac{3}{2 } \right ) } \ , \biggl [ 2\xi \bigl ( \frac{a_1}{a } \bigr)^{\!4 } - ( 1 \!-\ !
3\xi ) \bigl ( \frac{m}{h_i } \bigr)^{\!2 } \biggr ] \nonumber \\ & \hspace{0.5 cm } - \frac{\left ( \nu_i \!-\ ! \frac{1}{2 } \right ) \gamma(\nu_i ) } { 3\left ( \nu_i \!-\ ! \frac{3}{2 } \right ) } \bigl ( \frac{\mathcal{h}_0}{\mu } \bigr)^{\ ! 2\nu_i - 3 } \bigl ( \frac{a_1}{a } \bigr)^{\!4 } + \frac{2\sqrt{\pi}}{3 } ( \nu_i \!-\ !
1 ) \gamma\bigl ( \tfrac{3}{2 } \!-\ !
\nu_i \bigr ) \biggl [ ( 1 \!-\ ! 6\xi)(\nu_i\!-\!2)\bigl ( \frac{a_1}{a } \bigr)^{\ ! 4 } \nonumber \\ & \hspace{0.5 cm } - \frac { ( 1 \!-\ !
6\xi ) \left ( \nu_i \!-\ ! \frac{3}{2 }
\right ) + 3\nu_i \left ( \nu_i \!+\ ! \frac{3}{2 } \right ) + 2(1\!-\!6\xi)\nu_i^2 \left ( \nu_i \!+\ ! \frac{3}{2 } \right ) } { 6(\nu_i \!-\!1)\left ( \nu_i \!+\ ! \frac{1}{2 } \right ) \left ( \nu_i \!+\ ! \frac{3}{2 } \right ) } \bigl ( \frac{m}{h_i } \bigr)^{\!2 } \biggr ] \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{\ ! 3 \!-\ !
2\nu_i } \biggr\ } \ , .
\label{pqm}\end{aligned}\ ] ] we have given some parts of the expressions above in terms of physical quantities for the sake of clarity .
the comment of the dependence on the arbitrary cutoff @xmath149 is warranted .
it is bound to cancel with the same contribution from the remaining part of integration interval , and it does not contribute to the full result .
the reason we kept it explicitly is because we want to take the minimally coupled limit , which we do and discuss in the next subsection .
the minimally coupled limit consists in taking @xmath337 , and then expanding @xmath295 , defined in , for small mass in and , @xmath338 + \frac{h_i^4}{8\pi^2 } \biggl [ \ln\bigl ( \frac{a}{a_1 } \bigr ) - \frac{1}{2 } + \ln\bigl ( \frac{2\mu}{a_1 h_i } \bigr ) + \gamma_e \biggr ] \bigl ( \frac{a_1}{a } \bigr)^4 \ , , \label{rhoradmin } \\ p_q \approx { } & - \frac{3 h_i^4}{16\pi^2 } \biggl [ 1 - e^{- \frac{2}{3}\frac{m^2}{h_i^2 } n_i } \biggr ] + \frac{h_i^4}{24\pi^2 } \biggl [ \ln\bigl ( \frac{a}{a_1 } \bigr ) - \frac{3}{2 } + \ln\bigl ( \frac{2\mu}{a_1 h_i } \bigr ) + \gamma_e \biggr ] \bigl ( \frac{a_1}{a } \bigr)^4 \ , .
\label{pradmin}\end{aligned}\ ] ] note that the dependence on @xmath149 has persisted in the final answer .
this does not signal that the answer is wrong , but rather that the uv does not contribute a suppressed contribution . in the exactly massless limit of this result
the first terms drop out from the energy density and pressure above , and what remains is exactly the result from @xcite , where the massless minimally coupled case was studied from the start .
what effectively cuts off the uv radiation - like contribution is the finite time of transition @xmath48 between the inflationary and radiation period .
we have assumed here that the radiation period does not last excessively long , more precisely , @xmath339 where @xmath340 is the total number of e - foldings of radiation period , which will be satisfied by the requirements in the end .
this first terms in and coincide with the ones computed in @xcite , since they derive from @xmath341 term . note that this term is not the dominant one for a very small mass .
the leading order contribution in this limit is @xmath343 \ , , \label{radrholimit2 } \\
p_q \approx { } & -\frac{3 h_i^4}{32\pi^2 } \ , e^{8|\xi|n_i } \biggl [ \frac{1}{3 } \bigl ( \frac{a_1}{a } \bigr)^{\!4 } + \frac{1}{6|\xi| } \bigl ( \frac{m}{h_i } \bigr)^{\!2 } \biggr ] \ , , \label{radplimit2}\end{aligned}\ ] ] where we have assumed that the radiation period will not last longer ( in e - foldings ) than the inflationary period ( @xmath344 ) , which will be true in the end for the ranges of nonminimal couplings of interest in this work .
there are two qualitatively different contributions to the energy density and pressure above .
the first contribution is radiation - like and it redshifts just as the background does .
it is easy to see that if the constraint from figure [ inflconstraint ] is satisfied , this contribution will never dominate in radiation period .
the second contribution is of the cc type , and its energy density has a positive sign . for suitable choices of parameters one can get this contribution to be the dominant one in the backreaction during radiation period . and since it does not redshift away the ratio @xmath345 grows , and it might grow to order one for suitable masses and small enough nonminimal couplings .
but we are not interested in this scenario happening in radiation period . what we are interested in is realizing it in matter period , which we turn to next . during radiation period
we require the cc - type contribution to be negligible , in which case and reduce to the massless limit of @xcite , @xmath346 after the transition to the matter era , the following hierarchy of scales is reached ( fig .
[ hubble rate ] ) , @xmath347 where @xmath149 is a fiducial scale , introduced for the sake of isolating the relevant contribution to integrals , @xmath348 as discussed in section [ sec : energy density and pressure integrals ] .
the integrand here is @xmath349^ * \ , , \label{mattintegrand}\ ] ] where the bogolyubov coefficients are well approximated by the sudden transition ones and determined by two fast transitions from inflation to radiation , and from radiation to matter , @xmath350 u_m'^*(k,\eta_2 ) \nonumber \\ & \hspace{2 cm } + i \bigl [ \alpha_r(k ) u_r'(k,\eta_2 ) + \beta_r(k ) u_r'^*(k,\eta_2 ) \bigr ] u_m^*(k,\eta_2 ) \
\beta_m(k ) = { } & i \bigl [ \alpha_r(k ) u_r(k,\eta_2 ) + \beta_r(k ) u_r^*(k,\eta_2 ) \bigr ] u_m'(k,\eta_2 ) \nonumber \\ & \hspace{2 cm } + i \bigl [ \alpha_r(k )
u_r'(k,\eta_2 ) + \beta_r(k ) u_r'^*(k,\eta_2 ) \bigr ] u_m(k,\eta_2 ) \ , .\end{aligned}\ ] ] bogolyubov coefficients in radiation period @xmath351 and @xmath352 , appearing in the expression above , were already approximated in the previous subsection , where it was found that @xmath353 . applying this here
u_m'^*(k,\eta_2 ) + 2 \im
\bigl [ \beta_r^*(k ) u_r'(k,\eta_2 ) \bigr ] u_m^*(k,\eta_2 ) \ , , \\ \beta_m(k ) \approx { } & 2 \im\bigl [ \beta_r^*(k ) u_r(k,\eta_2 ) \bigr ] u_m'(k,\eta_2 ) - 2 \im \bigl [ \beta_r^*(k ) u_r'(k,\eta_2 ) \bigr ] u_m(k,\eta_2 ) \ , , \end{aligned}\ ] ] from where we see that again @xmath355
now the integrand simplifies to @xmath356 \biggr\}^2 \ , .\ ] ] which we can write out as @xmath357 = { } & 2 \im\bigl [ \beta_r^*(k )
u_r(k,\eta_2 ) \bigr ] \im\bigl [ u_m'^*(k,\eta_2 ) u_m(k,\eta ) \bigr ] \nonumber \\ & - 2 \im\bigl [ \beta_r^*(k ) u_r'(k,\eta_2 ) \bigr ] \im\bigl [ u_m^*(k,\eta_2 ) u_m(k,\eta ) \bigr ] \ , .\end{aligned}\ ] ] part of this integrand was already approximated in .
because of the hierarchy of scales in matter period , we may expand this further in the ir limit , @xmath358 \approx { } & - \frac{2^{\nu_i-2}}{\sqrt{\pi } } \gamma(\nu_i ) \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) \mathcal{h}_1^{\nu_i+1/2}\mathcal{h}_2^{-1 } k^{-\nu_i } \biggl [ 1 \!-\ ! \frac{1}{20 } \bigl ( \frac{ma_2}{\mathcal{h}_2 } \bigr)^{\!2 } \biggr ] \ , ,
\\ \im\bigl [ \beta_r^*(k ) u_r'(k,\eta_2 ) \bigr ] \approx { } & - \frac{2^{\nu_i-2}}{\sqrt{\pi } } \gamma(\nu_i ) \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr ) \mathcal{h}_1^{\nu_i+1/2 } k^{-\nu_i } \biggl [ 1 \!-\ ! \frac{1}{8 } \bigl ( \frac{ma_2}{\mathcal{h}_2 } \bigr)^{\!2 } \biggr ] \ , .\end{aligned}\ ] ] in the remaining part of the integrand we first use the property to express it solely in terms of mode functions and , @xmath359 = { } & \im\bigl [ a_m^*(k , m ) b_m(k , m ) \bigr ] \!\times\ ! \bigl [ v_{m1}(k,\eta_1 ) v_{m2}(k,\eta ) - v_{m2}(k,\eta_2 ) v_{m1}(k,\eta ) \bigr ] \nonumber \\ = { } & \frac{1}{2\nu } \bigl [ v_{m1}(k,\eta_1 ) v_{m2}(k,\eta ) - v_{m2}(k,\eta_2 ) v_{m1}(k,\eta ) \bigr ] \ , , \end{aligned}\ ] ] and then we use the hierarchy to simplify it further by expanding it to leading order in @xmath360 , @xmath361 \approx \frac{\gamma(\nu)}{2 } \mathcal{h}_2^{-1/2+\nu } \mathcal{h}^{-1/2 } k^{-\nu } \nonumber \\ & \hspace{1 cm } \times \biggl\ { j_\nu \bigl ( \frac{2k}{\mathcal{h } } \bigr ) + \bigl ( \frac{ma}{\mathcal{h } } \bigr)^2 \times \frac{\mathcal{h}}{30k } \biggl [ \bigl ( -6 \!+\ ! ( 1\!-\!\nu)(2\!-\!\nu ) \frac{\mathcal{h}^2}{k^2 } \bigr ) j_{1+\nu } \bigl ( \frac{2k}{\mathcal{h } } \bigr ) \nonumber \\ & \hspace{5.3 cm } + ( 2\!-\!\nu ) \bigl ( 3 \!-\ ! ( 1\!-\!\nu)(2\!+\!\nu ) \frac{\mathcal{h}^2}{k^2 } \bigr ) \frac{\mathcal{h}}{k } j_{2+\nu } \bigl ( \frac{2k}{\mathcal{h } } \bigr ) \biggr ] \biggr\ } \ , , \\ & \im\bigl [ u_m'^*(k,\eta_2 ) u_m(k,\eta ) \bigr ] \approx - \frac{1}{2 } \bigl ( \nu \!-\ ! \tfrac{1}{2 } \bigr ) \mathcal{h}_2 \times \im\bigl [ u_m^*(k,\eta_2 ) u_m(k,\eta ) \bigr ] \ , .\end{aligned}\ ] ] the full integrand is now given approximately as @xmath362 \biggr\}^2 \nonumber \\
= { } & \frac{2^{2\nu_i-4}}{\pi } \gamma^2(\nu_i ) \gamma^2(\nu ) \bigl ( \nu_i \!-\ ! \tfrac{1}{2 } \bigr)^2 \bigl ( \nu \!+\ ! \tfrac{3}{2 }
\bigr)^2 \mathcal{h}_1^{2\nu_i+1 } \mathcal{h}_2^{2\nu-1 } \mathcal{h}^{-1 } k^{-2\nu_i-2\nu } \nonumber \\ & \times \biggl\ { j_{\nu}^2\bigl ( \frac{2k}{\mathcal{h } } \bigr ) + \bigl ( \frac{ma}{\mathcal{h } } \bigr)^2 \times \frac{\mathcal{h}}{15k } j_\nu\bigl ( \frac{2k}{\mathcal{h } } \bigr ) \biggl [ \bigl ( - 6 \!+\ ! ( 1\!-\!\nu)(2\!-\!\nu ) \frac{\mathcal{h}^2}{k^2 } \bigr ) j_{1+\nu } \bigl ( \frac{2k}{\mathcal{h } } \bigr ) \nonumber \\ & \hspace{4 cm } + ( 2\!-\!\nu ) \bigl ( 3 \!-\ ! ( 1\!-\!\nu)(2\!+\!\nu)\frac{\mathcal{h}^2}{k^2 } \bigr ) \frac{\mathcal{h}}{k } j_{2+\nu}\bigl ( \frac{2k}{\mathcal{h } } \bigr ) \biggr ] \biggr\ } \ , .
\label{mattzapprox}\end{aligned}\ ] ] given the approximation , we can now perform integrals , and expand the result according to the hierarchy , @xmath363 \nonumber \\ & \hspace{1 cm } - \frac{2^{-3 + 2\nu+2\nu_i } \gamma\left ( \frac{3}{2}\!-\!\nu_i \right ) \gamma(\nu_i\!+\!\nu\!-\!1 ) } { \sqrt{\pi } \ , \gamma(\nu_i \!+\
! \nu \!-\ !
\frac{1}{2 } ) \gamma(\nu_i\!+\!2\nu\!-\!\frac{1}{2 } ) } \bigl ( \frac{\mathcal{h}_1}{\mathcal{h}_2 } \bigr)^{\!2\nu_i-3 } \bigl ( \frac{\mathcal{h}_2}{\mathcal{h } } \bigr)^{\!2\nu_i-3 } \bigl [ 1 + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } \bigr ] \biggr\ } \nonumber \\ & \times \bigl ( \frac{\mathcal{h}_1}{\mathcal{h}_2 } \bigr)^{\!4 } \bigl ( \frac{\mathcal{h}_2}{\mathcal{h } } \bigr)^{2\nu_i-3 } \mathcal{h}^2 \ , , \end{aligned}\ ] ] @xmath364 \nonumber \\ & \hspace{1 cm } + \frac{2^{-5 + 2\nu+2\nu_i } \gamma\left ( \frac{5}{2}\!-\!\nu_i \right ) \gamma(\nu_i\!+\!\nu\!-\!2 ) } { \sqrt{\pi } \ ,
\gamma(\nu_i \!+\ ! \nu \!-\ !
\frac{3}{2 } ) \gamma(\nu_i\!+\!2\nu\!-\!\frac{3}{2 } ) } \bigl ( \frac{\mathcal{h}_1}{\mathcal{h}_2 } \bigr)^{\!2\nu_i-3 } \bigl ( \frac{\mathcal{h}_2}{\mathcal{h } } \bigr)^{\!2\nu_i-3 } \mathcal{h}^2 \bigl [ 1 + \mathcal{o}\bigl ( \frac{ma}{\mathcal{h } } \bigr)^{\!2 } \bigr ] \biggr\ } \nonumber \\ & \times \bigl ( \frac{\mathcal{h}_1}{\mathcal{h}_2 } \bigr)^{\!4 } \bigl ( \frac{\mathcal{h}_2}{\mathcal{h } } \bigr)^{2\nu_i-3 } \mathcal{h}^2 \ , .\end{aligned}\ ] ] plugging these two integrals into and yields the backreaction energy density and pressure in matter era , @xmath365 \bigl ( \frac{a_2}{a } \bigr)^{\frac{3}{2}-\nu } \nonumber \\ & - \frac{(1\!-\!6\xi)(\nu_i\!+\!\nu\!-\!3)\ , \gamma\!\left ( \frac{3}{2}\!-\!\nu_i \right ) \gamma^2(\nu ) \ , \gamma(2\nu_i\!+\!2\nu\!-\!4 ) } { \gamma^2\!\left ( \nu_i \!+\ ! \nu\!-\!\frac{3}{2 } \right ) \gamma\!\left ( \nu_i \!+\ ! 2\nu \!-\ !
\frac{1}{2 } \right ) } \ , e^{-4n_r } \bigl(\frac{a_2}{a } \bigr)^{\!\frac{9}{2}-\nu } \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{2\nu_i-3 } \biggr\ } \ , , \label{mattrhofull}\end{aligned}\ ] ] @xmath366 \bigl ( \frac{a_2}{a } \bigr)^{\frac{3}{2}-\nu } \nonumber \\ & + \frac{(1\!-\!6\xi)(\nu_i\!+\!\nu\!-\!3)^2\ , \gamma\!\left ( \frac{3}{2}\!-\!\nu_i \right ) \gamma^2(\nu ) \ , \gamma(2\nu_i\!+\!2\nu\!-\!4 ) } { 3\,\gamma^2\!\left ( \nu_i \!+\ ! \nu\!-\!\frac{3}{2 } \right ) \gamma\!\left ( \nu_i \!+\ ! 2\nu \!-\ !
\frac{1}{2 } \right ) } \ , e^{-4n_r } \bigl(\frac{a_2}{a } \bigr)^{\!\frac{9}{2}-\nu } \bigl ( \frac{\mathcal{h}_0}{\mathcal{h } } \bigr)^{2\nu_i-3 } \biggr\ } \ , .
\label{mattpfull}\end{aligned}\ ] ] the terms containing factors @xmath367 must be negligible in order for us to have control over the approximation .
the reason behind this lies in the ir regularization employed in this computation .
we have introduced a sharp ir cutoff @xmath299 in comoving momentum space arguing that it is an approximation of the contribution coming from a full state ( from all scales ) that is smoothly suppressed in the deep ir .
the deep ir scales are assumed to contribute subdominantly , and hence they were dropped right from the start by the introduction of this cutoff .
but once @xmath368 is reached by the cosmological evolution it means that there are no more modes in the ir except the ones below the cutoff scale @xmath96 , so they are the only ones contributing relevantly to the backreaction ( this was treated in the massless limit in @xcite ) .
this is where our approximation breaks down .
the physical requirement that we make on the model is that the inflation lasts long enough for the expressions above to be reliable .
this will indeed be true for the cases of interest in this work . setting @xmath337 in and produces the minimally coupled limit , @xmath369 \ , , \label{mattrhomin } \\
p_q \approx { } & - \frac{3h_i^4}{16\pi^2 } \biggl [ 1 - e^{-\frac{2}{3 } \frac{m^2}{h_i^2}n_i } \biggr ] \ , .
\label{mattpmin}\end{aligned}\ ] ] there are two types of contributions to the backreaction energy density and pressure here . the first one scales like nonrelativistic matter , and redshifts at the same rate as the background fluid driving the expansion .
it is a contribution that was found in the massless minimally coupled case @xcite .
the second contribution is of the cc type , making it interesting in the context of de scenarios .
there are two limits one can discuss regarding the second term , which we do in the following . in the limit of very long inflation , @xmath370 ,
the cc - type contribution saturates to a maximum value during inflation , and remains constant throughout expansion ( provided @xmath371 always ) .
it is easy to see that it contributes dominantly to and ( since it does not redshift away ) , @xmath372 this is a scenario that was suggested in @xcite , although the rigorous computation has not been performed , which we supply in this paper .
it was found that in order for this limit to work as a de model ( requiring @xmath373 ) , _ i.e. _ for the backreaction to have the right value at late - time matter era , one must considerably lower the inflationary hubble scale @xmath374 ( corresponding to energy scale of inflation @xmath375 ) . the conditions @xmath376 for very long inflation , and @xmath377 then imply that @xmath5 . in the limit of `` short '' inflation , @xmath378 , the cc - type contribution in and does not have enough time to reach its maximum value in inflation , but the value it has at the end of inflation freezes throughout subsequent expansion ( provided @xmath371 always ) . for short enough inflation
the backreaction energy density and pressure at late times in matter era are @xmath379 the matter - like contribution never becomes important @xcite compared to the background , and that is why we have neglected it above .
this is the result and the scenario suggested in @xcite .
it was found that in order to work as a de scenario , one does not have to lower the inflationary scale , @xmath380 , but still a very long inflation is required , @xmath381 ( together with the mass being lighter than the hubble rate today , @xmath382 ) .
the computations of the quantum backreaction in @xcite were performed for inflationary and radiation periods , but a rigorous computation for the matter period was missing ( even though the predicted result was correct ) , which is supplied by the limit taken in this subsection . in this limit
the leading contributions to the quantum backreaction energy density and pressure are @xmath383 \ , , \label{matter era : rhoq } \\
p_q \approx { } & - \frac{3h_i^4}{32\pi^2 } \ , e^{8|\xi|n_i } \biggl [ \frac{1}{6|\xi| } \bigl ( \frac{m}{h_i } \bigr)^{\!2 } \biggr ] \ , .
\label{matter era : pq}\end{aligned}\ ] ] one can recognize two types of contributions first one that survives in the massless limit and scales like nonrelativistic matter , and another that behaves like the cc and depends on the mass .
we want to discuss in which cases the cc - type contribution dominates the backreaction , and under which conditions can it be large enough to influence the background dynamics .
firstly , since the first matter - like contribution behaves like a tracer solution in matter and radiation era ( see and ) , its ratio compared to the background is determined by the ratio at the end of inflation , which then freezes for subsequent evolution .
so , if the conditions of figure [ inflconstraint ] are met , if the backreaction is small at the end of inflation ( @xmath384 ) , this term will never be important .
the second cc - type contribution does not redshift away and can become comparable to the background , under the condition that @xmath385 ( otherwise the field becomes heavy and starts contributing like nonrelativistic matter ) .
a more precise constraint on the scalar field mass is to compare it to the hubble rate at the onset of de domination which we take to be the time of equality of the cc and matter defined by @xmath386 .
it follows quickly from the first friedmann equation and the density parameters of cc and matter today , @xmath387 and @xmath388 @xcite , that @xmath389 corresponding to @xmath390 e - foldings in the past of today ( redshift @xmath391 ) .
the measure of the strength of the backreaction at late times is the ratio , @xmath392 when this becomes of order one the backreaction starts dominating .
up until that moment the quantum backreaction behaves like a cosmological constant with positive energy density , so its tendency would be to speed up the expansion rate .
the exact details of this process are a matter of performing the full self - consistent evolution , but at least the initial tendency is clear .
this model has three parameters @xmath54 , @xmath1 , and @xmath306 that are not completely independent .
the mass is constrained to be smaller than the hubble rate at the start of de domination , @xmath393 the nonminimal coupling @xmath1 and the duration of inflation @xmath306 are constrained in fig .
[ inflconstraint ] by the requirement that backreaction remains perturbative during inflation .
the crucial requirement for the model to work is that the ratio is of order one .
this condition determines the number of e - foldings as a function of the nonminimal coupling @xmath1 and the ratio , @xmath394 \ , , \label{nipredicted}\ ] ] where @xmath316 is the planck energy .
the plot of @xmath395 for the case @xmath396 is shown in fig .
[ nimin ] .
part of the curve in fig .
[ nimin ] does not lie in the allowed region of fig .
[ inflconstraint ] , and hence the model does not work for the whole considered range of nonminimal couplings @xmath397 , but still @xmath398 is allowed .
of course , the limits on nonminimal coupling depend on .
in fact , we can derive this bound on nonminimal coupling dependent on the ratio @xmath399 by requiring that the predicted number of e - foldings satisfies the constraint from inflation , which gives @xmath400 which does not depend on the inflationary scale @xmath307 . since for the case depicted in fig .
[ nimin ] @xmath396 , the results should be trusted to the 10% level due to possible subleading mass corrections .
+ 0.3 cm and the total number of e - foldings of inflation @xmath306 determined by condition @xmath312 , and @xmath396 , represented by the bold curve .
the shaded region is excluded by the requirements that quantum backreaction stays perturbative until late times ( fig .
[ inflconstraint]).,title="fig:",width=377 ] -0.5 cm
in this paper we investigate the evolution of quantum fluctuations of a very light , non - minimally coupled , spectator scalar field ( [ action : massive nonminimal scalar ] ) throughout the history of the universe , from the beginning of inflation and throughout inflation , radiation and matter era . when the field couples to a fixed classical homogeneous and isotropic cosmological background ( characterised by the hubble rate as a function of time , @xmath401 ) , and when gravity is assumed non - dynamical ( _ i.e _ the quantum gravitational effects are turned off ) , the relevant scalar field equation ( [ klein - gordon equation:1 ] ) is linear and can thus in principle be solved exactly , at least with the help of numerical methods .
since here we are interested in the ( one - loop quantum ) backreaction of the scalar quantum fluctuations on the background space - time , numerically solving our problem turns out to be tedious ( but can be done as in @xcite ) . instead , here we resort to approximate analytical treatment . indeed , as illustrated in fig .
[ expansion history ] , it is convenient to split the history of the universe into relatively long epochs , during which the principal slow roll parameter , @xmath402 , is to a good approximation constant , and relatively short transition periods , during which @xmath40 rapidly changes . we show that , provided the transition periods are sufficiently short ( for the transition in question that means that the characteristic time scale must be shorter than the hubble time at the transition ) , they can be treated in a sudden transition approximation .
this is so because the main contribution to the energy - momentum tensor , that determines the backreaction on the background space - time , comes from the infrared modes for which the sudden transition approximation applies .
it is important to investigate how the results depend on the choice of initial state ( which when chosen naively suffers from ir divergences ) , and the ir regularization method .
we have argued in subsection [ subsec : choice of state ] that the three viable regularization method must give at least qualitatively the same answer , which was supported by comparisons of explicit computations done using different schemes .
therefore , the results of our analysis are _ quite generic _ , _ i.e. _ to a large extent independent on the choice of the initial state .
it is also important to emphasize that , in the process of calculating the one - loop energy - momentum tensor , we have used dimensional regularization to remove all divergences and that our final result for the renormalized energy - momentum tensor is finite and cutoff independent .
the cutoff independence is not trivial to achieve , since in the process of dimensional regularization one has to introduce an ultraviolet cutoff splitting the uv and ir parts .
regularization and renormalization procedures for the uv part are outlined in appendix [ app : regularization and renormalization ] .
the cutoff dependence introduced this way is only fiducial and cancels completely when the two parts are added up , resulting in a cutoff independent , fully regulated , finite answer .
it is now a good moment to state the most important results of this work .
the one - loop late - time contribution to the energy density and pressure in matter era , in the limit when @xmath403 and @xmath404 are of the form ( see ( [ matter era : rhoq][matter era : pq ] ) ) , @xmath405 where @xmath54 is the scalar field mass , and @xmath1 its nonminimal coupling to the ricci scalar ( see action ) , @xmath406 is the inflationary hubble rate .
note that seemingly the one - loop energy density and pressure in are proportional to @xmath407 .
this is a consequence of the convention for writing the quadratic part of the potential in terms of particle mass @xmath54 . but
@xmath408 , which appears in action instead of @xmath54 when dimensionful units are restored , should be considered as constant , and not as singular in the classical limit @xmath409 .
this means is proportional to @xmath410 in the usual one - loop sense .
since @xmath411 is approximately constant and @xmath412 , this contribution can be _ a good candidate for dark energy_. the principal goal of the upcoming work @xcite in which we study in the gaussian approximation the self - consistent one - loop backreaction in the model ( [ action : massive nonminimal scalar ] ) is to establish whether this nave proposition is justified .
namely , the perturbative treatment employed in this work fails to be reliable when the backreaction becomes significant , which is precisely when it becomes interesting from the dark energy point of view .
the contribution ( [ summary : rhoq : matter era:1 ] ) can be compared with the energy density at the onset of de domination , @xmath413 ( where @xmath414 was defined in ) , to give , @xmath415 where the last factor represents the enhancement factor due to the inflationary particle production . in section [ matter era ]
we show ( see eq .
( [ condition on xi ] ) ) that the expression ( [ summary : rhoq : matter era:2 ] ) is valid provided the following inequalities are satisfied , @xmath416 the enhancement factor @xmath417 in eq .
( [ summary : rhoq : matter era:2 ] ) needs to be sufficiently large to compensate the loop suppression factor , @xmath418 \!=\ !
g_nh_i^2\hbar/[3\pi c^5 ] \!\sim\
! 10^{-13}$ ] , and the factor @xmath419 ^ 2 $ ] .
the number of e - foldings of inflation required for that follows immediately from ( [ summary : rhoq : matter era:2 ] ) , @xmath420 \ , , \label{number of efolds required}\ ] ] where @xmath421 is the planck energy .
this number of e - foldings , also shown in fig .
[ nimin ] , must be consistent with the requirement that particle production does not lead to dominant contribution to the energy density during inflation , _
i.e. _ must be consistent with the bound shown in fig .
[ inflconstraint ] .
albeit these two conditions significantly restrict the allowed parameter space for @xmath422 , one can show that , provided @xmath423 and @xmath424 , there still exists an ample set of allowed choices for @xmath1 . in the following
we discuss how our result ( [ number of efolds required ] ) for the required number of e - foldings in our model with negative nonminimal coupling @xmath1 compares with other models , in particular with the minimally coupled case ( when @xmath337 ) studied in ref . @xcite . in the minimally coupled case ,
the amplification factor in is simply @xmath306 , such that @xmath425 while the number of e - foldings implied by eq .
( [ number of efolds required ] ) is typically hundreds or thousands , the number of e - foldings implied by the minimally coupled case ( [ number of efolds required:2 ] ) is of the order or greater than @xmath6 ( for @xmath426 ) , which is many orders of magnitude larger .
there is a simple explanation for this large difference : while the rate of particle production in the minimally coupled case generates secular effects in the one - loop energy - momentum tensor that grow linearly in time , due to the tachyonic scalar mass generated by a negative @xmath1 during inflation particle production rate in the nonminimally coupled case generates an exponentially growing contribution to the one - loop energy - momentum tensor .
this , at first sight small difference in the model , has thus very important consequences for model building and arguably favors scalar models with negative non - minimal coupling as the model of choice for dark energy from inflationary quantum fluctuations .
this paper is not the first to investigate the possibility that vacuum fluctuations of matter fields may be responsible for dark energy and here we make a cursory overview of these earlier works .
apart from our earlier papers @xcite , in which we investigated the late - time one - loop quantum backreaction from one - loop inflationary fluctuations of massless and minimally and nonminimally coupled scalars , and of gravitons , probably the closest to our work are the papers of ringeval , suyama , takahashi , yamaguchi and yokoyama @xcite , of aoki and iso @xcite , results of which we discussed in the previous paragraph , and of parker and raval @xcite , and of parker and vanzella @xcite .
@xcite investigated the late - time one - loop quantum backreaction from one - loop inflationary fluctuations of very light minimally coupled scalars .
the conclusion of ref .
@xcite is that , provided inflation lasts long enough and it is at the right scale , the model can account for the observed dark energy , while the conclusion of ref .
@xcite ( which improves on @xcite by performing a computation similar to the one presented in this work ) is that inflation can occur at the grand - unified scale and provided it lasts for about @xmath6 e - foldings ( see ( [ number of efolds required:2 ] ) ) scalar field fluctuations can account for the dark energy .
however , in @xcite , a careful removal of all cut - off dependences , and a careful construction of approximate mode functions in matter era are not accounted for , which we properly include in this paper . in a series of papers that appeared soon after the original supernovae results , parker and raval @xcite used the effective gravitational action obtained by integrating out the matter fields .
the method used was the schwinger - dewitt proper time method and @xmath427 function regularization @xcite , which is an ultraviolet expansion of the effective action ( that holds at short geodesic distances ) , and the final effective action is presented as an expansion in powers of @xmath428 , where @xmath429 is the effective mass of the field ( in the case of a non - minimally coupled scalar , @xmath430 , where @xmath54 is the tree - level scalar mass and @xmath431 denotes the ricci scalar curvature ) . in their effective action parker and raval maintain the terms of the order @xmath432 , @xmath433 , @xmath434 , @xmath435 , @xmath436 , and @xmath437 , thus neglecting the inverse powers of @xmath433 . strictly speaking ,
this expansion applies ( and hence the truncation is reasonable ) when @xmath438 , where @xmath439 denote a suitably chosen norm
. a careful look at those papers reveals that the analysis was conducted strictly speaking where the expansion does not apply , _
i.e. _ in the region of parameter space where @xmath433 is of the order or smaller than the components of e ricci tensor , making the conclusions questionable ( in fact abundant particle production in the parker - raval model occurs when @xmath440 , which is precisely where @xmath441 , at which point the expansion used to construct the model is unreliable ) .
in contrast , the analysis in this work is performed in the opposite regime , namely in the regime when @xmath442 .
this is not just a technical point , but an essential assumption required to get abundant particle production during inflation that we need in order to get large quantum backreaction discussed in this paper . making a more detailed comparison with the first paper
@xcite reveals that a large late - time backreaction was obtained when @xmath443 , _ i.e. _ when @xmath1 is close to , but smaller than conformal coupling , @xmath444 .
this is to be contrasted with our results , which indicate that a large quantum backreaction is obtained only when both conditions , ( 1 ) @xmath423 and ( 2 ) there is a sufficiently long inflationary period preceding radiation era , are satisfied .
furthermore , the quantum backreaction in @xcite becomes large ( during matter era ) at a particular redshift @xmath445 given by @xmath446 = m^2\omega_{m0}/[3h_0 ^ 2(\tfrac{1}{6}\!-\!\xi)]$ ] ( here @xmath447 denotes the matter density today and @xmath448 \!\simeq\ ! 0.3 $ ] and @xmath449 is the hubble parameter today ) , at which moment a large particle production occurs due to an instability .
we see no sign of this kind of instability , albeit in fairness to the reader we note that our analysis is perturbative and therefore we might not be able to see such an instability . in the remaining papers @xcite the same effective action is used , and hence the same comparative analysis applies . the results presented in this work provide a very good motivation for further studies of the quantum backreaction effects in connection to the dark energy problem . in particular , the model presented here invites for more detailed studies that would : * treat the backreaction self - consistently , * confront the model with the currently existing dark energy data , * make forecasts that would test the model against the upcoming data , * examine the clustering properties of dark energy in this model , and * address the issue of a very light scalar field mass required for the model to work . by self - consistent treatment we mean solving self - consistently the quantum - corrected friedmann equations ( with the one - loop backreaction included ) together with the scalar field equations .
the perturbative treatment executed in this work fails as soon as the one - loop backreaction terms become comparable with the background contributions @xmath450 and @xmath451 .
the idea of the upcoming work @xcite is to extend the starobinsky s stochastic formalism for inflation to subsequent epochs and solve the resulting scalar field equations together with the friedmann equations ( that include the backreaction from the field fluctuations ) .
this will allow us to get detailed predictions on how the ( global ) hubble parameter depends on time when the backreaction starts dominating , and study its dependence on the principal parameters of our model : @xmath54 , @xmath1 and @xmath306 .
of course the choice of these parameters is already quite limited by the constraints discussed at length in this work .
the results of the self - consistent study can then be used to confront the model with the existing ( and upcoming ) data that already today put rather strict constraints on the recent evolution of dark energy . while in the initial work in ref . @xcite we intend to study the dynamics of the universe in the gaussian approximation , in the follow up
work we intend to generalize that work and calculate ( perturbatively ) the higher order ( non - gaussian ) correlators , which can be used to further test the model .
namely , one of the hot topics in current studies of dark energy models involves the question whether dark energy clusters and if it does , how much .
we expect that our model makes very specific predictions on how large non - gaussian features in dark energy are today and how they evolve in recent times .
next , we would like to make theoretical improvements of the model discussed in this work .
in order for our model to work , the scalar field mass has to be very light .
more precisely , it has to satisfy ( [ ultralight mass and small xi ] ) , which means that @xmath54 has to be smaller than the hubble scale at the onset of de domination , @xmath452 .
unless one has a mechanism for generation of such a tiny mass scale , this very light scalar mass remains a mystery of the model . in future work
we intend to investigate ( dynamical ) mechanisms that can shed light on the question why the scalar mass is so tiny today . to conclude , in light of the results presented in this work it is clear that the possibility that inflationary quantum fluctuations could be responsible for the observed dark energy should be taken seriously
. however , it is also clear that much more work is needed to put that idea on more solid foundations .
t.t . would like to thank institute for theoretical physics at utrecht university for the hospitality during the visit , where a part of this work has been done .
the work of t.t . is partially supported by jsps kakenhi grant number 15k05084 and mext kakenhi grant number 15h05888 .
this work is part of the d - itp consortium , a program of the netherlands organisation for scientific research ( nwo ) that is funded by the dutch ministry of education , culture and science ( ocw ) . during the final stages of the project d.g .
was supported by the grant 2014/14/e / st9/00152 ( national science centre , poland ) .
this appendix summarizes the dimensional regularization of the energy - momentum tensor integrals and , and the renormalization procedure by introducing counterterms .
it is comprised of well known results @xcite .
the first task is to isolate the divergences in and . in order to do that we examine just the uv parts of the integrals , @xmath453 \mathcal{h } ' |u|^2 \nonumber \\ & + 2m^2a^2 |u|^2 - \frac{1}{2 } \bigl [ d \!-\ ! 2 \!-\ !
4\xi(d \!-\ !
1 ) \bigr ] \mathcal{h } \frac{\partial}{\partial\eta } |u|^2 + \frac{1}{2 } \frac{\partial^2}{\partial\eta^2 } |u|^2 \biggr\ } \ , , \label{rhouv } \\
p_q^{\text{uv},0 } = { } & \frac{\delta_{ij } a^{-d}}{(4\pi)^{\frac{d-1}{2 } } \ , \gamma \ !
\left ( \frac{d-1}{2 } \right ) } \int\limits_{\mu}^{\infty } \!\ !
dk \ , k^{d-2 } \biggl\ { \frac{2k^2}{(d \!-\ !
1 ) } |u|^2 - \frac{1}{2 } \bigl [ d \!-\ ! 2 \!-\ !
4\xi(d \!-\ !
1 ) \bigr ] \mathcal{h } ' |u|^2 \nonumber \\ & - \frac{1}{2 } \bigl [ d \!-\ !
4\xi(d \!-\ !
1 ) \bigr ] \mathcal{h } \frac{\partial}{\partial\eta } |u|^2 + \frac{(1 \!-\ ! 4\xi)}{2 } \frac{\partial^2}{\partial\eta^2 } |u|^2 \biggr\ } \ , , \label{puv}\end{aligned}\ ] ] where the uv scale @xmath149 is assumed to be larger than any other physical scale ( such as the curvature and the mass of the scalar ) .
the momentum in the integrands is then much larger than any other physical scale appearing , whic is exploited in the following subsection .
therefore , we do not need the exact momentum dependence in order to isolate the divergences , its asymptotic expansion in powers of @xmath114 ( where @xmath454 ) is enough in order to accomplish the task .
for that reason we first solve the equation of motion , @xmath75 u(k,\eta ) = 0 \ , , \qquad k\rightarrow\infty \ , , \label{eomapp}\ ] ] in the uv limit .
the wkb method is well suited for that .
here we implement a somewhat simpler method , which is directly expanding the solution in inverse powers of momenta , @xmath455 \ , , \label{uansatzuv}\ ] ] where we have assumed the mode function to be a pure positive - frequency one in the uv .
the coefficient functions @xmath456 are all real , which does not seem to be a general ansatz , but it can be shown that their imaginary parts can always be absorbed into a time - independent phase , which we can always add since it does not show up in physical quantities .
it is enough to determine @xmath457-@xmath458 for the purpose of isolating uv divergences in and .
equation of motion they satisfy are obtained by plugging ansatz into , and organizing it order by order in momenta , @xmath459 \ , , \label{eom2 } \\
u_3'(\eta ) = { } & - \frac{1}{2 } \bigl [ u_2''(\eta ) + \mathcal{m}^2(\eta ) u_2(\eta ) \bigr ] \
, , \label{eom3 } \\ u_4'(\eta ) = { } & \frac{1}{2 } \bigl [ u_3''(\eta ) + \mathcal{m}^2(\eta ) u_3(\eta ) \bigr ] \ , . \label{eom4}\end{aligned}\ ] ] we also have to impose the wronskian normalization , which we do order by order as well , @xmath460 the solutions to equations of motion - , and constraints and are , @xmath461 \ , , \\
u_2(\eta ) = { } & - \frac{\mathcal{m}^2(\eta)}{4 } - \frac{1}{8}\biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^2(\tau ) \biggr]^2 \ , , \\
u_3(\eta ) = { } & \frac{1}{8 } \bigl ( [ \mathcal{m}^2(\eta ) ] ' - [ \mathcal{m}^2(\eta_0 ) ] ' \bigr ) + \frac{\mathcal{m}^2(\eta)}{8 } \biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^2(\tau ) \biggr ] \nonumber \\ & + \frac{1}{48 } \biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^2(\tau ) \biggr]^3 + \frac{1}{8 } \biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^4(\tau ) \biggr ] \\
u_4(\eta ) = { } & \frac{[\mathcal{m}^2(\eta)]''}{16 } + \frac{5\mathcal{m}^4(\eta)}{32 } + \frac{1}{16 } \bigl ( [ \mathcal{m}^2(\eta ) ] ' - [ \mathcal{m}^2(\eta_0 ) ] ' \bigr ) \biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^2(\tau ) \biggr ] \nonumber \\ & + \frac{\mathcal{m}^2(\eta)}{32 } \biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^2(\tau ) \biggr]^2 + \frac{1}{384 } \biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^2(\tau ) \biggr]^4 \nonumber \\ & + \frac{1}{16 } \biggl[\int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^2(\tau ) \biggr ] \biggl [ \int_{\eta_0}^{\eta}\ ! d\tau \mathcal{m}^4(\tau ) \biggr ] \ , .\end{aligned}\ ] ] the dependence on an arbitrary time @xmath43 ( which corresponds to integration constants from - ) is physically irrelevant in the sense that it can be absorbed into the time - independent overall phase which has no physical meaning . in fact , what we need to calculate and is just the modulus squared , @xmath462 + \frac{1}{k^4 } \bigl [ 2u_4(\eta ) + 2u_1(\eta ) u_3(\eta ) + u_2 ^ 2(\eta ) \bigr ] + \mathcal{o}(k^{-6 } ) \biggr\ } \ , , \ ] ] which is independent of all the integration constants , @xmath463 '' \bigr ] + \mathcal{o}(k^{-6 } ) \biggr\ } \ , .
\label{u2expansion}\ ] ] now it is straightforward to evaluate integrals and , @xmath464 \biggr ) \nonumber \\ & + \frac{\mu^{d-4}}{(d \!-\ !
4 ) } \times \frac{a^{-d}}{(4\pi)^{\frac{d-1}{2 } } \gamma\left ( \frac{d-1}{2 } \right ) } \biggl\ { \frac{(ma)^4}{8 } + \frac{(d \!-\ !
6)}{16 } \bigl [ d \!-\ ! 2 \!-\
! 4\xi(d \!-\ !
1 ) \bigr ] ( ma)^2 \mathcal{h}^2 \nonumber \\ & + \frac{1}{128 } \bigl [ d \!-\ ! 2 \!-\ ! 4\xi(d \!-\ !
1 ) \bigr]^2 \bigl [ 8 \mathcal{h}''\mathcal{h } - 4(\mathcal{h}')^2 - 3(d \!-\ !
2)^2\mathcal{h}^4 \bigr ] \biggr\ } + \mathcal{o}(\mu^{-2 } ) \ , , \label{rhouv0}\end{aligned}\ ] ] @xmath465 \biggr ) \nonumber \\ & + \frac{\mu^{d-4}}{(d \!-\ !
4 ) } \times \frac{a^{-d}}{(4\pi)^{\frac{d-1}{2}(d\!-\!1 ) } \gamma\left ( \frac{d-1}{2 } \right ) } \biggl\ { -\frac{3(ma)^4}{8 } - \frac{(d \!-\ !
6)}{16 } \bigl [ d \!-\ ! 2 \!-\ !
4\xi(d \!-\ !
1 ) \bigr ] ( ma)^2 ( 2\mathcal{h } ' \!+\ !
\mathcal{h}^2 ) \nonumber \\ & + \frac{1}{128 } \bigl [ d \!-\ ! 2 \!-\ ! 4\xi(d \!-\ !
1 ) \bigr]^2 \bigl [ -8\mathcal{h } ' '' + 8 \mathcal{h}''\mathcal{h } - 4(\mathcal{h}')^2 \nonumber \\ & \hspace{3 cm } + 12(d \!-\ !
2)^2\mathcal{h}'\mathcal{h}^2 -3(d\!-\!2)^2 \mathcal{h}^4 \bigr ] \biggr\ } + \mathcal{o}(\mu^{-2 } ) \ , .
\label{puv0}\end{aligned}\ ] ] the action for the counterterms , which include a cosmological constant one , a newton s constant one , and a higher - derivative one , is given by @xmath466 from where its their contribution to the energy - momentum tensor follows , @xmath467 where @xmath468 specialized to flrw space - time the counterterms contribution to the energy - momentum tensor is diagonal , @xmath469 \ , , \label{rhoct } \\
p_{\text{ct } } = { } & - \frac{\delta\lambda}{8\pi g_{\!n } } + \frac{\delta g_{\!n}^{-1}}{16\pi } \frac{(d \!-\ !
2)}{a^2 }
\bigl [ 2 \mathcal{h } ' + ( d \!-\ ! 3)\mathcal{h}^2 \bigr ] \nonumber \\ & + ( \alpha \!+\ !
\delta\alpha ) \frac{(d \!-\ ! 1)}{a^4 } \bigl [ 8 \mathcal{h } ' '' + 8(2d \!-\ !
9)\mathcal{h}''\mathcal{h } + 4(3d \!-\ !
11)(\mathcal{h}')^2 \nonumber \\ & \hspace{3 cm } + 12(d^2 \!-\ !
10d \!+\ ! 20 ) \mathcal{h}'\mathcal{h}^2 + ( d \!-\ ! 2)(d \!-\ ! 5)(d \!-\ !
10 ) \mathcal{h}^4 \bigr ] \ , . \label{pct}\end{aligned}\ ] ] we choose the coefficients of the counterterms to be @xmath470m^2 } { 8 ( 4\pi)^{\frac{d-3}{2 } } ( d\!-\!2 ) \ , \gamma\!\left ( \frac{d-1}{2 } \right ) } \frac{\mu^{d-4}}{(d \!-\ !
4 ) } \ , , \\ \delta\alpha = { } & \frac{[d \!-\ ! 1 \!-\
! 4\xi(d \!-\ !
1)]^2 } { 128(d \!-\ ! 1)^2 ( 4\pi)^{\frac{d-1}{2 } } \ , \gamma\!\left ( \frac{d-1}{2 } \right ) } \frac{\mu^{d-4}}{(d \!-\ !
4 ) } \ , .\end{aligned}\ ] ] in order to absorb the divergences from and .
the finite parts of these coefficients were picked for convenience so as to cancel as much of finite parts of the bare expectation value . the renormalized uv contribution to energy density and pressure
is obtained by adding the counterterms contribution and to the bare contribution and , @xmath471 \biggr\ } - \frac{m^4}{32\pi^2 } \ln(a ) + \frac{(1 \!-\ !
6\xi ) m^2}{16\pi^2a^2 } \mathcal{h}^2 \ln(a ) \nonumber \\ & + \frac{(1 \!-\
! 6\xi)^2}{32\pi^2 a^4 } \biggl\ { \bigl [ \ln(a)+ \widetilde{\alpha } \bigr ] \bigl [ -2\mathcal{h}''\mathcal{h } \!+\ !
( \mathcal{h}')^2 \!+\ !
3 \mathcal{h}^4 \bigr ] -2 ( \mathcal{h}'\mathcal{h}^2 \!+\ !
\mathcal{h}^4 ) \biggr\ } \ , , \label{rhoquv}\end{aligned}\ ] ] @xmath472 \biggr\ } \nonumber \\ & + \frac{m^4}{32\pi^2 } \bigl [ \ln(a ) + \frac{1}{3 } \ , \bigr ] - \frac{(1 \!-\ !
6\xi ) m^2}{48\pi^2a^2 } \bigl [ ( 2\mathcal{h } ' \!+\ !
\mathcal{h}^2 ) \ln(a ) + \mathcal{h}^2 \bigr ] \nonumber \\ & + \frac{(1 \!-\ !
6\xi)^2}{96\pi^2a^4 } \biggl\ { \bigl [ \ln(a ) + \widetilde{\alpha } \bigr ] \bigl [ 2\mathcal{h } ' '' \!-\
! 2\mathcal{h}''\mathcal{h }
( \mathcal{h}')^2 \!-\ ! 12\mathcal{h}'\mathcal{h}^2 \!+\ !
3\mathcal{h}^4 \bigr ] \nonumber \\ & \hspace{2 cm } + 4\mathcal{h}''\mathcal{h } \!+\ !
3(\mathcal{h}')^2 \!+\ ! 6 \mathcal{h}'\mathcal{h}^2 \!-\
! 5\mathcal{h}^4 \biggr\ } \ , , \label{pquv}\end{aligned}\ ] ] where @xmath473 .
specialized to constant @xmath40 flrw backgrounds these contributions are @xmath471 \biggr\ } - \frac{m^4}{32\pi^2 } \ln(a ) + \frac{(1 \!-\ !
6\xi ) m^2}{16\pi^2a^2 } \mathcal{h}^2 \ln(a ) \nonumber \\ & + \frac{(1 \!-\ !
6\xi)^2}{32\pi^2 } ( 2 \!-\ !
\epsilon ) \frac{\mathcal{h}^4}{a^4 } \biggl\ { 3 \epsilon\bigl [ \ln(a)+ \widetilde{\alpha } \bigr ] - 2 \biggr\ } \
, , \label{rhouvconstepsilon}\end{aligned}\ ] ] @xmath474 \biggr\ } + \frac{m^4}{32\pi^2 } \bigl [ \ln(a ) + \frac{1}{3 } \ , \bigr ] \nonumber \\ & - \frac{(1 \!-\ !
6\xi ) m^2}{48\pi^2 } \frac{\mathcal{h}^2}{a^2 } \bigl [ ( 3 \!-\ !
2\epsilon ) \ln(a ) + 1 \bigr ] \nonumber \\ & + \frac{(1 \!-\ ! 6\xi)^2}{96\pi^2 } ( 2 \!-\ !
\epsilon ) \frac{\mathcal{h}^4}{a^4 } \biggl\ { -3\epsilon(3 \!-\ !
4\epsilon ) \bigl [ \ln(a ) + \widetilde{\alpha } \bigr ] + ( 6 \!-\ !
11\epsilon ) \biggr\ } \ , .
\label{puvconstepsilon}\end{aligned}\ ] ] had we performed the renormalization procedure on arbitrary curved backgrounds , we would have found another contribution that breaks the classical conformal invariance the conformal anomaly @xcite , @xmath475 \ , , \\ p_{\text{ca } } = { } & - \frac{1}{8640\pi^2a^4 } \bigl [ 2\mathcal{h } ' '' \!-\ !
2\mathcal{h}''\mathcal{h } \!+\ ! ( \mathcal{h}')^2 \bigr ] \ , , \label{ca}\end{aligned}\ ] ] which should be added to the contributions above .
specialized to constant @xmath40 backgrounds it is @xmath476 a. g. riess _ et al . _
[ supernova search team collaboration ] , `` observational evidence from supernovae for an accelerating universe and a cosmological constant , '' astron .
j. * 116 * ( 1998 ) 1009 [ astro - ph/9805201 ] .
d. glavan , t. prokopec and v. prymidis , `` backreaction of a massless minimally coupled scalar field from inflationary quantum fluctuations , '' phys .
d * 89 * ( 2014 ) 2 , 024024 [ arxiv:1308.5954 [ gr - qc ] ] .
d. glavan , t. prokopec and d. c. van der woude , `` late - time quantum backreaction from inflationary fluctuations of a nonminimally coupled massless scalar , '' phys .
d * 91 * ( 2015 ) 2 , 024014 [ arxiv:1408.4705 [ gr - qc ] ] .
f. finelli , g. marozzi , a. a. starobinsky , g. p. vacca and g. venturi , `` generation of fluctuations during inflation : comparison of stochastic and field - theoretic approaches , '' phys .
d * 79 * ( 2009 ) 044007 [ arxiv:0808.1786 [ hep - th ] ] .
f. finelli , g. marozzi , a. a. starobinsky , g. p. vacca and g. venturi , `` stochastic growth of quantum fluctuations during slow - roll inflation , '' phys .
d * 82 * ( 2010 ) 064020 [ arxiv:1003.1327 [ hep - th ] ] . c. ringeval , t. suyama , t. takahashi , m. yamaguchi and s. yokoyama , `` dark energy from primordial inflationary quantum fluctuations , '' phys .
* 105 * ( 2010 ) 121301 [ arxiv:1006.0368 [ astro-ph.co ] ] .
l. parker , `` quantized fields and particle creation in expanding universes . 1 .
, '' phys .
* 183 * ( 1969 ) 1057 .
l. parker , `` quantized fields and particle creation in expanding universes . 2 . , ''
d * 3 * ( 1971 ) 346 [ phys .
d * 3 * ( 1971 ) 2546 ] .
y. b. zeldovich and a. a. starobinsky , `` particle production and vacuum polarization in an anisotropic gravitational field , '' sov .
jetp * 34 * ( 1972 ) 1159 [ zh .
* 61 * ( 1971 ) 2161 ]
. t. m. janssen and t. prokopec , `` regulating the infrared by mode matching : a massless scalar in expanding spaces with constant deceleration , '' phys .
d * 83 * ( 2011 ) 084035 [ arxiv:0906.0666 [ gr - qc ] ] .
l. parker , w. komp and d. vanzella , `` cosmological acceleration through transition to constant scalar c curvature , '' astrophys .
j. * 588 * ( 2003 ) 663 [ astro - ph/0206488 ] .
l. parker and d. a. t. vanzella , `` acceleration of the universe , vacuum metamorphosis , and the large time asymptotic form of the heat kernel , '' phys .
d * 69 * ( 2004 ) 104009 [ gr - qc/0312108 ] .
l. parker and d. j. toms , `` new form for the coincidence limit of the feynman propagator , or heat kernel , in curved space - time , '' phys .
d * 31 * ( 1985 ) 953 .
i. jack and l. parker , `` proof of summed form of proper time expansion for propagator in curved space - time , '' phys .
d * 31 * ( 1985 ) 2439 . | we investigate the backreaction of the quantum fluctuations of a very light ( @xmath0 ) nonminimally coupled spectator scalar field on the expansion dynamics of the universe .
the one - loop expectation value of the energy momentum tensor of these fluctuations , as a measure of the backreaction , is computed throughout the expansion history from the early inflationary universe until the onset of recent acceleration today .
we show that , when the nonminimal coupling @xmath1 to ricci curvature is negative ( @xmath2 corresponding to conformal coupling ) , the quantum backreaction grows exponentially during inflation , such that it can grow large enough rather quickly ( within a few hundred e - foldings ) to survive until late time and constitute a contribution of the cosmological constant type of the right magnitude to appreciably alter the expansion dynamics .
the unique feature of this model is in that , under rather generic assumptions , inflation provides natural explanation for the initial conditions needed to explain the late - time accelerated expansion of the universe , making it a particularly attractive model of dark energy . |
the duluth systems in senseval-2 take a supervised learning approach to the spanish and english lexical sample tasks .
they learn decision trees and naive bayesian classifiers from sense - tagged training examples where the context in which an ambiguous word occurs is represented by lexical features . these include unigrams and bigrams that occur anywhere in the context , and co - occurrences within just a few words of the target word .
these are the only types of features used .
there are no syntactic features , nor is the structure or content of wordnet employed . as a result
these systems are highly portable , and can serve as a foundation for systems that are tailored to particular languages and sense inventories .
the word sense disambiguation literature provides ample evidence that many different kinds of features contribute to the resolution of word meaning .
these include part of speech , morphology , verb object relationships , selectional restrictions , lexical features , etc .
when used in combination it is often unclear to what degree each type of feature contributes to overall performance .
it is also unclear to what extent adding new features allows for the disambiguation of previously unresolvable test instances .
one of the long term objectives of our research is to determine which types of features are complementary and cover increasing numbers of test instances as they are added to a representation of context .
the training and test data for the english and spanish lexical sample tasks is split into separate training and test files per word .
a supervised learning algorithm induces a classifier from the training examples for a word , which is then used to assign sense tags to the test instances for that word .
the context in which an ambiguous word occurs is represented by lexical features that are identified using the bigram statistics package ( bsp ) version 0.4 . this is free software that extracts unigrams and bigrams from text using a variety of statistical methods .
each unigram or bigram that is identified in the training data is treated as a binary feature that indicates whether or not it occurs in the context of the word being disambiguated .
the free software package sensetools ( version 0.1 ) converts training and test data into a feature vector representation , based on the output from bsp .
this becomes the input to the weka suite of supervised learning algorithms .
weka induces classifiers from the training examples and applies the sense tags to the test instances .
the same software is used for the english and spanish text .
bsp and sensetools are written in perl and are freely available from www.d.umn.edu/~tpederse/code.html .
weka is written in java and is freely available from www.cs.waikato.ac.nz/~ml .
there were eight pairs of duluth systems in the english and spanish lexical sample tasks .
the only language dependent components are the tokenizers and stop lists . for both english and spanish a stop list
is made up of all words that occur ten or more times in five randomly selected word training files of comparable size .
all duluth systems exclude the words in the stop
list from being features .
each pair of systems is summarized below .
all performance results are based on accuracy ( correct / total ) using fine - grained scoring .
the name of the english system appears first , followed by the spanish system .
* duluth1/duluth6 * create an ensemble of three naive bayesian classifiers , where each is based on a different set of features .
the hope is that these different views of the training examples will result in classifiers that make complementary errors , and that their combined performance will be better than any of the individual classifiers .
separate naive bayesian classifiers are learned from each representation of the training examples .
each classifier assigns probabilities to each of the possible senses of a test instance .
these are summed and the sense with the largest value is used .
this technique is used in many of our ensembles and will be referred to as a weighted vote .
the first feature set is made up of bigrams , i.e. , consecutive two word sequences , that can occur anywhere in the context with the ambiguous word . to be selected as a feature ,
a bigram must occur two or more times in the training examples and have a log - likelihood ratio ( @xmath0 ) value @xmath1 6.635 , which is associated with a p - value of .01 .
the second feature set is based on unigrams , i.e. , one word sequences , that occur five or more times in the training data .
the third feature set is made up of co - occurrence features that represent words that occur on the immediate left or right of the target word . in effect , these are bigrams that include the target word .
they must also occur two or more times and have a log - likelihood ratio @xmath1 2.706 , which is associated with a p - value of .10 .
these systems are inspired by @xcite , which presents an ensemble of eighty - one naive bayesian classifiers based on varying sized windows of context to the left and right of the target word that define co - occurrence features .
however , the current systems only use a three member ensemble to capture the spirit of simplicity and portability that underlies the duluth approach to senseval-2 .
english accuracy was 53% , spanish was 58% .
* learn an ensemble of decision trees via bagging .
ten samples are drawn , with replacement , from the training examples for a word .
a decision tree is learned from each of these permutations of the training examples , and each of these trees becomes a member of the ensemble . a test instance is assigned a sense based on a weighted vote among the members of the ensemble . in general decision tree learning
can be overly influenced by a small percentage of the training examples , so the goal of bagging is to smooth out this instability .
there is only one kind of feature used in these systems , bigrams that occur two or more times and have a log - likelihood ratio @xmath1 6.635 .
this is one of the three feature sets used in the duluth1/duluth6 systems .
the set of bigrams that meet these criteria become candidate features for the j48 decision tree learning algorithm , which is the weka implementation of the c4.5 algorithm .
the decision tree learner first constructs a tree of features that characterizes the training data exactly , and then prunes features away to avoid over fitting and allow it to generalize to the previously unseen test instances .
thus , a decision tree learner performs a second cycle of feature selection and is not likely to use all of the features that we identify prior to learning with bsp .
the default c4.5 parameter settings are used for pruning .
these systems are an extension of @xcite , which learns a single decision tree where the representation of context is based on bigrams .
this earlier work does not use bagging , and the top 100 bigrams according to the log - likelihood ratio are the candidate features .
english accuracy was 54% , spanish was 60% .
* duluth3/duluth8 * rely on the same features as duluth1/duluth6 , but learn an ensemble of three bagged decision trees instead of an ensemble of naive bayesian classifiers .
there is a strong contrast between these techniques , since decision tree learners attempt to characterize the training examples and find relationships among the features , while a naive bayesian classifier is based on an assumption of conditional independence among the features .
the feature set used in these systems is from duluth1/duluth6 and consists of bigrams , unigrams and co - occurrences .
a bagged decision tree is learned for each of the three kinds of features .
the test instances are classified by each of the bagged decision trees , and a majority vote is taken among the members to assign senses to the test instances .
these are the most accurate of the duluth systems for both english ( 57% ) and spanish ( 61% ) .
these are within 7% of the most accurate overall approaches for english ( 64% ) and spanish ( 68% ) .
* duluth4/duluth9 * uses a naive bayesian classifier based on a bag of words representation of context , where each unigram that occurs in the training data is taken as a feature .
this is a common benchmark in word sense disambiguation studies and text classification problems . in the english training examples
any word that occurs five or more times is used as a feature , and in the spanish data any word that occurs two or more times is used .
these features are used to estimate the parameters of a naive bayesian classifier .
this will assign the most probable sense to a test instance , given the surrounding context .
accuracy for english was 54% , and for spanish 56% .
this naive bayesian classifier was one of the three member classifiers in the ensemble approach of duluth1/duluth7 , which was 1% less accurate for english and and 2% more accurate for spanish .
* duluth5/duluth10 * add a co occurrence feature to the duluth2/duluth7 systems . in every other respect
they are identical .
the co - occurrence feature was also used in duluth1/duluth6 , and is essentially a bigram where one of the words is the ambiguous word .
these must occur two or more times in the training examples and have a log - likelihood ratio @xmath1 2.706 to be included as a feature .
in addition to the co - occurrence feature the bigram feature from duluth2/duluth7 is used , where a bigram must occur two or more times and have a log - likelihood ratio @xmath1 6.635 .
accuracy for english was 55% , and for spanish 61% .
this was a slight improvement over duluth2 ( 54% ) and duluth7 ( 60% ) . * dulutha / duluthx * build an ensemble of three different classifiers that are induced from the same representation of the training examples . a weighted vote is taken to assign senses to test instances .
the three classifiers are a bagged j48 decision tree , a naive bayesian classifier , and the nearest neighbor classifier ib@xmath2 , where the number of neighbors parameter @xmath2 is set to 1 .
the context in which the ambiguous word occurs is represented by bigrams that may include zero , one , or two intervening words that are ignored . to be considered as features these bigrams must occur two or more times and have a log - likelihood ratio @xmath1 10.827 , i.e. , a p - value of .001 .
the log - likelihood ratio threshold is set to 0 for the spanish data due to the smaller volume of data .
english accuracy was 52% , spanish was 58% . * duluthb / duluthy * are identical to duluth5/duluth10 , except that rather than learning an entire decision tree they stop the learning process once the root of the decision tree is selected .
the resulting one node decision tree is called a decision stump . at worst
a decision stump will reproduce the most common sense baseline , and may do better if the selected feature is particularly informative . in previous work we have observed that decision stumps can serve as a very aggressive lower bound on performance @xcite .
decision stumps are the least accurate method for both english ( duluthb , 51% ) and spanish ( duluthy , 52% ) , but are more accurate than the most common sense baseline for english ( 48% ) and spanish ( 47% ) . * duluthc / duluthz * take a kitchen sink approach to ensemble creation , and combine the seven systems for english and spanish into ensembles that assign senses to test instances by taking a weighted vote among the members .
accuracy for english was 55% , and for spanish 59% .
this is less than the accuracy of some of the members systems , suggesting that the members of the ensemble are making redundant errors .
there are several hypotheses that underly and motivate these systems .
this hypothesis is at the core of much of our recent work .
it holds that variations in learning algorithms matter far less to disambiguation performance than do variations in the features used to represent the context in which an ambiguous word occurs . in other words
, an informative feature set will result in accurate disambiguation when used with a wide range of learning algorithms , but there is no learning algorithm that can overcome the limitations of an uninformative or misleading set of features .
there are a number of demonstrations that can be made from the duluth systems in support of this hypothesis , but perhaps the clearest is found in comparing the systems duluth1/duluth6 and duluth3/duluth8 .
the first pair learns three naive bayesian classifiers and the second learns three bagged decision trees .
both use the same feature set to represent the context in which ambiguous words occur .
there is a 3% improvement in accuracy when using the decision trees .
we believe this modest improvement when moving from a simple learning algorithm to a more complex one supports the hypothesis that the true dividends are to be found in improving the feature set .
we hypothesize a 50/25/25 rule for supervised approaches to word sense disambiguation .
this loosely holds that given a classifier learned from a sample of sense tagged training examples , about half of the test instances are easily disambiguated , a quarter are harder but still possible , and the remaining quarter are extremely difficult .
this is a minor variant of the 80/20 rule of time management , which holds that 20% of effort accounts for 80% of results . when the two highest ranking systems in the official english lexical sample results are compared there are 2180 test instances ( 50% ) that both disambiguate correctly using fine - grained scoring
there are an additional 1183 instances ( 28% ) where one of the two systems are correct , and 965 instances ( 22% ) that neither system can resolve .
if these two systems were optimally combined , their accuracy would be 78% .
if the third - place system is also considered , there are 1939 instances ( 44.8% ) that all three systems can disambiguate , and 816 ( 19% ) that none could resolve . for all the duluth systems for english , there are 1705 instances ( 39% ) that all eight systems got correct .
there are 1299 instances ( 30% ) that none can resolve
. the accuracy of an optimally combined system would be 70% .
the most accurate individual system is duluth3 with 57% accuracy . for the spanish duluth systems
, there are 856 instances ( 38% ) that all eight systems got correct
. there are 478 instances ( 21% ) that none of the systems got correct .
this results in an optimally combined result of 79% .
the most accurate duluth system was duluth8 , with 1369 correct instances ( 62% ) .
if the top ranked spanish system ( 68% ) and duluth8 are compared , there are 1086 instances ( 49% ) where both are correct , 737 instances ( 33% ) where one or the other is correct , and 402 instances ( 18% ) where neither system is correct .
this is intended as a rule of thumb , and suggests that a fairly substantial percentage of test instances can be resolved by almost any means , and that a hard core of test instances will be very difficult for any method to resolve .
this work has been partially supported by a national science foundation faculty early career development award ( # 0092784 ) . the bigram statistics package and sensetools
have been implemented by satanjeev banerjee .
t. pedersen .
a simple approach to building ensembles of naive bayesian classifiers for word sense disambiguation . in _ proceedings of the first annual meeting of the north american chapter of the association for computational linguistics _
, pages 6369 , seattle , wa , may . t. pedersen .
a decision tree of bigrams is an accurate predictor of word sense . in _ proceedings of the second annual meeting of the north american chapter of the association for computational linguistics _ , pages 7986 , pittsburgh , july . | this paper describes the sixteen duluth entries in the senseval-2 comparative exercise among word sense disambiguation systems .
there were eight pairs of duluth systems entered in the spanish and english lexical sample tasks .
these are all based on standard machine learning algorithms that induce classifiers from sense - tagged training text where the context in which ambiguous words occur are represented by simple lexical features .
these are highly portable , robust methods that can serve as a foundation for more tailored approaches . |
cluster - tilted algebras were introduced by buan , marsh and reiten @xcite and , independently in @xcite for type @xmath0 as a byproduct of the now extensive theory of cluster algebras of fomin and zelevinsky @xcite . since then , cluster - tilted algebras have been the subject of several investigations , see , for instance , @xcite . in particular , in @xcite
is given a construction procedure for cluster - tilted algebras : let @xmath1 be a triangular algebra of global dimension two over an algebraically closed field @xmath2 , and consider the @xmath1-@xmath1-bimodule @xmath3 , where @xmath4 is the standard duality , with its natural left and right @xmath1-actions .
the trivial extension of @xmath1 by this bimodule is called the _ relation - extension _
@xmath5 of @xmath1 .
it is shown there that , if @xmath1 is tilted , then its relation - extension is cluster - tilted , and every cluster - tilted algebra occurs in this way .
our purpose in this paper is to study the relation - extensions of a wider class of triangular algebras of global dimension two , namely the class of quasi - tilted algebras , introduced by happel , reiten and smal in @xcite . in general , the relation - extension of a quasi - tilted algebra is not cluster - tilted , however it is 2-calabi - yau tilted , see theorem [ thm 2.1 ] below .
we then look more closely at those cluster - tilted algebras which are tame and representation - infinite . according to @xcite ,
these coincide exactly with the cluster - tilted algebras of euclidean type .
we ask then the following question : given a cluster - tilted algebra @xmath6 of euclidean type , find all quasi - tilted algebras @xmath1 such that @xmath7 . a similar question has been asked ( and answered ) in @xcite ,
where , however , @xmath1 was assumed to be tilted . for this purpose ,
we generalize the notion of reflections of @xcite .
we prove that this operation allows to produce all tilted algebras @xmath1 such that @xmath7 , see theorem [ thm mainreflection ] . in @xcite
this result was shown only for cluster - tilted algebras of tree type .
we also prove that , unlike those of @xcite , reflections in the sense of the present paper are always defined , that the reflection of a tilted algebra is also tilted of the same type , and that they have the same relation - extension , see theorem [ thm reflection ] and proposition [ prop reflection ] below .
because all tilted algebras having a given cluster - tilted algebra as relation - extension are given by iterated reflections , this gives an algorithmic answer to our question above .
after that , we look at the tubes of a cluster - tilted algebra of euclidean type and give a procedure for constructing those tubes which contain a projective , see proposition [ prop5 ] .
we then return to quasi - tilted algebras in our last section , namely we define a particular two - sided ideal of a cluster - tilted algebra , which we call the partition ideal .
our first result ( theorem [ thm ctaqt ] ) shows that the quasi - tilted algebras which are not tilted but have a given cluster - tilted algebra @xmath6 of euclidean type as relation - extension are the quotients of @xmath6 by a partition ideal .
we end the paper with the proof of our main result ( theorem [ thm ctaqt3 ] ) which says that if @xmath1 is quasi - tilted and such that @xmath7 , then either @xmath1 is the quotient of @xmath6 by the annihilator of a local slice ( and then @xmath1 is tilted ) or it is the quotient of @xmath6 by a partition ideal ( and then @xmath1 is not tilted except in two cases easy to characterize ) .
throughout this paper , algebras are basic and connected finite dimensional algebras over a fixed algebraically closed field @xmath2 . for an algebra @xmath1 ,
we denote by @xmath8 the category of finitely generated right @xmath1-modules .
all subcategories are full , and identified with their object classes . given a category @xmath9 , we sometimes write @xmath10 to express that @xmath11 is an object in @xmath9 . if @xmath9 is a full subcategory of @xmath8 , we denote by @xmath12 the full subcategory of @xmath8 having as objects the finite direct sums of summands of modules in @xmath9 . for a point @xmath13 in the ordinary quiver of a given algebra @xmath1
, we denote by @xmath14 , @xmath15 , @xmath16 respectively , the indecomposable projective , injective and simple @xmath1-modules corresponding to @xmath13 .
we denote by @xmath17 the auslander - reiten quiver of @xmath1 and by @xmath18 the auslander - reiten translations . for further definitions and facts , we refer the reader to @xcite .
let @xmath19 be a finite connected and acyclic quiver .
a module @xmath20 over the path algebra @xmath21 of @xmath19 is called _ tilting _ if @xmath22 and the number of isoclasses ( isomorphism classes ) of indecomposable summands of @xmath20 equals @xmath23 , see @xcite .
an algebra @xmath1 is called _ tilted of type @xmath19 _ if there exists a tilting @xmath21-module @xmath20 such that @xmath24 .
it is shown in @xcite that an algebra @xmath1 is tilted if and only if it contains a _ complete slice _
@xmath25 , that is , a finite set of indecomposable modules such that * @xmath26 is a sincere @xmath1-module .
* if @xmath27 is a sequence of nonzero morphisms between indecomposable modules with @xmath28 then @xmath29 for all @xmath30 ( _ convexity _ ) .
* if @xmath31 is an almost split sequence in @xmath8 and at least one indecomposable summand of @xmath11 lies in @xmath25 , then exactly one of @xmath32 belongs to @xmath25 . for more on tilting and tilted algebras , we refer the reader to @xcite .
tilting can also be done within the framework of a hereditary category .
let @xmath33 be an abelian @xmath2-category which is hom - finite , that is , such that , for all @xmath34 , the vector space @xmath35 is finite dimensional .
we say that @xmath33 is _ hereditary _ if @xmath36 .
an object @xmath37 is called a _ tilting object _ if @xmath38 and the number of isoclasses of indecomposable objects of @xmath20 is the rank of the grothendieck group @xmath39 .
the endomorphism algebras of tilting objects in hereditary categories are called _ quasi - tilted algebras_. for instance , tilted algebras but also canonical algebras ( see @xcite ) are quasi - tilted .
quasi - tilted algebras have attracted a lot of attention and played an important role in representation theory , see for instance @xcite .
let @xmath19 be a finite , connected and acyclic quiver .
the _ cluster category _
@xmath40 of @xmath19 is defined as follows , see @xcite .
let @xmath41 denote the composition @xmath42 $ ] , where @xmath43 denotes the inverse auslander - reiten translation in the bounded derived category @xmath44 , and [ 1 ] denotes the shift of @xmath45 .
then @xmath40 is the orbit category @xmath46 : its objects are the @xmath41-orbits @xmath47 of the objects @xmath48 , and the space of morphisms from @xmath47 to @xmath49 is @xmath50 then @xmath40 is a triangulated category with almost split triangles and , moreover , for @xmath51 we have a bifunctorial isomorphism @xmath52 .
this is expressed by saying that the category @xmath40 is _ 2-calabi - yau_. an object @xmath53 is called _ tilting _ if @xmath54 and the number of isoclasses of indecomposable summands of @xmath55 equals @xmath23 .
the endomorphism algebra @xmath56 is then called _ cluster - tilted _ of type @xmath19 . more generally , the endomorphism algebra @xmath57 of a tilting object @xmath55 in a @xmath58-calabi - yau category with finite dimensional hom - spaces is called a _
2-calabi - yau tilted algebra _ , see @xcite .
let now @xmath20 be a tilting @xmath21-module , and @xmath24 the corresponding tilted algebra .
then it is shown in @xcite that the trivial extension @xmath59 of @xmath1 by the @xmath1-@xmath1-bimodule @xmath60 with the two natural actions of @xmath1 , the so - called _ relation - extension _ of @xmath1 , is cluster - tilted .
conversely , if @xmath6 is cluster - tilted , then there exists a tilted algebra @xmath1 such that @xmath61 .
let now @xmath6 be a cluster - tilted algebra , then a full subquiver @xmath25 of @xmath62 is a _ local slice _
, see @xcite , if : * @xmath25 is a _ presection _ , that is , if @xmath63 is an arrow then : * * @xmath64 implies that either @xmath65 or @xmath66 * * @xmath65 implies that either @xmath67 or @xmath68 .
* @xmath25 is _ sectionally convex _ , that is , if @xmath69 is a sectional path in @xmath62 then @xmath70 implies that @xmath71 for all @xmath30 . *
@xmath72 .
let @xmath1 be tilted , then , under the standard embedding @xmath73 , any complete slice in the tilted algebra @xmath1 embeds as a local slice in @xmath74 , and any local slice in @xmath75 occurs in this way . if @xmath6 is a cluster - tilted algebra , then a tilted algebra @xmath1 is such that @xmath61 if and only if there exists a local slice @xmath25 in @xmath62 such that @xmath76 , where @xmath77 , see @xcite .
let @xmath25 be a local slice in the transjective component of @xmath62 having the property that all the sources in @xmath25 are injective @xmath6-modules .
then @xmath25 is called a _ rightmost _ slice of @xmath6 .
let @xmath13 be a point in the quiver of @xmath6 such that @xmath15 is an injective source of the rightmost slice @xmath25 .
then @xmath13 is called a _ strong sink_. _ leftmost slices _ and _ strong sources _ are defined dually .
we start with a motivating example .
let @xmath1 be the tilted algebra of type @xmath78 given by the quiver @xmath79_{\beta}\\1&&4\ar[lu]_{\alpha}\ar[ld]^{\gamma}\\&3\ar[lu]^{\delta}}\ ] ] bound by @xmath80 , @xmath81 .
its relation - extension is the cluster - tilted algebra @xmath6 given by the quiver @xmath82_{\beta}\\1\ar@<1.5pt>[rr]^{\lambda}\ar@<-1.5pt>[rr]_\mu&&4\ar[lu]_{\alpha}\ar[ld]^{\gamma}\\&3\ar[lu]^{\delta}}\ ] ] bound by @xmath80 , @xmath83 , @xmath84 , @xmath81 , @xmath85 , @xmath86 .
however , @xmath6 is also the relation - extension of the algebra @xmath87 given by the quiver @xmath88_{\alpha}&1\ar@<1.5pt>[l ] _ { ^{\lambda}}\ar@<-1.5pt>[l]^{\ \atop\mu}&3\ar[l]_{\delta}}\ ] ] bound by @xmath84 , @xmath85 . this latter algebra @xmath87 is not tilted , but it is quasi - tilted . in particular , it is triangular of global dimension two . therefore , the question arises natrually whether the relation - extension of a quasi - tilted algebra is always cluster - tilted .
this is certainly not true in general , for the relation - extension of a tubular algebra is not cluster - tilted .
however , it is 2-calabi - yau tilted . in this section , we prove that the relation - extension of a quasi - tilted algebra is always 2-calabi - yau tilted .
let @xmath89 be a hereditary category with tilting object @xmath20 .
because of @xcite , there exist an algebra @xmath90 , which is hereditary or canonical , and a triangle equivalence @xmath91 .
let @xmath92 denote the image of @xmath20 under this equivalence .
because @xmath93 preserves the shift and the auslander - reiten translation , it induces an equivalence between the cluster categories @xmath94 and @xmath95 , see ( * ? ? ?
* section 4.1 ) . indeed , because @xmath90 is canonical or hereditary , it follows that @xmath96 , where @xmath97.$ ] therefore , we have @xmath98 .
we say that a 2-calabi - yau tilted algebra @xmath99 is of _ canonical type _ if the 2-calabi - yau category @xmath100 is the cluster category of a canonical algebra .
the proof of the next theorem follows closely @xcite .
[ thm 2.1 ] let @xmath1 be a quasi - tilted algebra .
then its relation - extension @xmath59 is cluster - tilted or it is 2-calabi - yau titled of canonical type . because @xmath1 is quasi - tilted , there exist a hereditary category @xmath89 and a tilting object @xmath20 in @xmath89 such that @xmath101 . as observed above , there exist an algebra @xmath90 , which is hereditary or canonical , and a triangle equivalence @xmath91 .
let @xmath102.we have @xmath103 , and therefore @xmath104 , c[2 ] ) \\ & \cong & { \textup{hom}}_{{\mathcal{d}^b(\mathcal{h})}}(\tau t[1 ] , t[2 ] ) \\ & \cong & { \textup{hom}}_{{\mathcal{d}^b(\mathcal{h } ) } } ( t , \tau^{-1 } t[1 ] ) \\ & \cong & { \textup{hom}}_{{\mathcal{d}^b(\mathcal{h } ) } } ( t , f t ) .\\ \end{array}\ ] ] thus the additive structure of @xmath105 is that of @xmath106 then , we check exactly as in ( * ? ? ? * section 3.3 ) that the multiplicative structure is preserved .
this completes the proof .
let @xmath1 be a representation - infinite quasi - tilted algebra .
then @xmath1 is derived equivalent to a hereditary or a canonical algebra @xmath90 .
let @xmath107 denote the tubular type of @xmath90 .
we then say that @xmath1 has canonical type @xmath108 .
[ lem 1 ] let @xmath1 be a representation - infinite quasi - tilted .
then its relation - extension @xmath5 is cluster - tilted of euclidean type if and only if @xmath109 is one of @xmath110 indeed , @xmath5 is cluster - tilted of euclidean type if and only if @xmath1 is derived equivalent to a tilted algebra of euclidean type , and this is the case if and only if @xmath109 belongs to the above list .
[ rem 2 ] it is possible that @xmath1 is domestic , but yet @xmath5 is wild .
indeed , we modify the example after corollary d in @xcite .
recall from @xcite that there exists a tame concealed full convex subcategory @xmath111 such that @xmath1 is a semiregular branch enlargement of @xmath111 @xmath112k[f_j],\ ] ] where @xmath113 are ( truncated ) branches . then the representation theory of @xmath1 is determined by those of @xmath114k$ ] and @xmath115 $ ] .
let @xmath1 be given by the quiver @xmath116_{\delta}&11\ar[l]_\zeta\\ & 3\ar[ul]_{\alpha}\ar[dl]^{\beta}&4\ar[l]_{\gamma}\ar[d]^\nu&5\ar[l]_{\sigma}\\ 2&&8\ar[d]^\varphi & 9\ar[l]_\omega&7\ar[ul]_\rho \\&&10 } \ ] ] bound by the relations @xmath117 , @xmath118 , @xmath119 . here
@xmath120 is the full subcategory generated by @xmath121 and @xmath122 the one generated by @xmath123 .
then @xmath120 has domestic tubular type @xmath124 and @xmath122 has domestic tubular type @xmath125 .
therefore @xmath1 is domestic . on the other hand ,
the canonical type of @xmath1 is @xmath126 , which is wild . in this example , the 2-calabi - yau tilted algebra @xmath5 is not cluster - tilted , because it is not of euclidean type , but the derived category of @xmath127 contains tubes , see @xcite .
there clearly exist algebras which are not quasi - tilted but whose relation - extension is cluster - tilted of euclidean type .
indeed , let @xmath1 be given by the quiver @xmath128^{\alpha}&5\ar[r]^{\beta}&4\ar[r]^{\gamma}&3\ar[r]^{\delta}&2\ar@<-2pt>[r]_\mu\ar@<2pt>[r]^{\lambda}&1}\ ] ] bound by @xmath129 .
then @xmath1 is iterated tilted of type @xmath78 of global dimension 2 , see @xcite .
its relation - extension is given by @xmath128^{\alpha}&5\ar[r]^{\beta}&4\ar@/_25pt/[ll]_{\sigma}\ar[r]^{\gamma}&3\ar[r]^{\delta}&2\ar@<-2pt>[r]_\mu\ar@<2pt>[r]^{\lambda}&1\ar@/_25pt/[ll]_\eta}\ ] ] bound by @xmath130 .
this algebra is isomorphic to the relation - extension of the tilted algebra of type @xmath78 given by the quiver @xmath131^{\sigma}\ar[r]^{\gamma}&3\ar[r]^{\delta}&2\ar@<-2pt>[r]_\mu\ar@<2pt>[r]^{\lambda}&1\\ 5\ar[ru]^{\beta}}\ ] ] bound by @xmath132 , @xmath133 .
therefore @xmath5 is cluster - tilted of euclidean type . on the other hand , @xmath1 is not quasi - tilted , because the uniserial module @xmath134 has both projective and injective dimension 2 .
let @xmath1 be a tilted algebra .
let @xmath135 be a rightmost slice , and let @xmath15 be an injective source of @xmath135 .
thus @xmath13 is a strong sink in @xmath1 .
we define the _ completion @xmath136 of @xmath13 _ by the following three conditions .
* @xmath137 . * @xmath136 is closed under predecessors in @xmath135 . *
if @xmath138 is an arrow in @xmath135 with @xmath139 having an injective successor in @xmath136 then @xmath140 . observe that @xmath136 may be constructed inductively in the following way .
we let @xmath141 , and @xmath142 be the closure of @xmath143 with respect to ( c ) ( that is , we simply add the direct successors of @xmath15 in @xmath135 , and if a direct successor of @xmath15 is injective , we also take its direct successor , etc . )
we then let @xmath144 be the closure of @xmath142 with respect to predecessors in @xmath135 .
then we repeat the procedure ; given @xmath145 , we let @xmath146 be the closure of @xmath145 with respect to ( c ) and @xmath147 be the closure of @xmath146 with respect to predecessors
. this procedure must stabilize , because the slice @xmath135 is finite .
if @xmath148 with @xmath149 , we let @xmath150
. we can decompose @xmath136 as the disjoint union of three sets as follows .
let @xmath151 denote the set of injectives in @xmath136 , let @xmath152 be the set of non - injectives in @xmath136 which have an injective successor in @xmath136 , and let @xmath153 denote the complement of @xmath154 in @xmath136 . thus @xmath155 is a disjoint union .
[ rem h ] if @xmath156 then @xmath136 reduces to the completion @xmath157 as defined in @xcite . recall that @xmath157 does not always exist , but , as seen above , @xmath136 does .
conversely , if @xmath157 exists , then it follows from its construction in @xcite that @xmath156 .
thus @xmath156 if and only if @xmath157 exists , and , in this case @xmath158 . for every module @xmath11 over a cluster - tilted algebra @xmath6
, we can consider a lift @xmath159 in the cluster category @xmath100 .
abusing notation , we sometimes write @xmath160 to denote the image of @xmath161 in @xmath162 , and say that the auslander - reiten translation is computed in the cluster category .
let @xmath13 be a strong sink in @xmath1 and let @xmath135 be a rightmost local slice with injective source @xmath15 . recall that @xmath135 is also a local slice in @xmath162 .
then the reflection of the slice @xmath135 in @xmath13 is @xmath163 where @xmath164 is computed in the cluster category . in a similar way , one defines the coreflection @xmath165 of leftmost slices with projective sink @xmath166 .
[ thm reflection ] let @xmath13 be a strong sink in @xmath1 and let @xmath135 be a rightmost local slice in @xmath162 with injective source @xmath15 . then the reflection @xmath167 is a local slice as well .
set @xmath168 and @xmath169 where again , @xmath170 and @xmath164 are computed in the cluster category @xmath100 .
we claim that @xmath170 is a local slice in @xmath100 .
notice that since @xmath136 is closed under predecessors in @xmath135 , then , if @xmath171 is a neighbor of @xmath172 , we must have an arrow @xmath173 in @xmath135 .
this observation being made , @xmath170 is clearly obtained from @xmath135 by applying a sequence of apr - tilts .
thus @xmath170 is a local slice in @xmath100 .
we now claim that @xmath174 is closed under predecessors in @xmath170 .
indeed , let @xmath175 and @xmath176 be such that we have an arrow @xmath173 .
then , there exists an arrow @xmath177 in the cluster category . because @xmath178 , we have @xmath179 .
now if @xmath180 , then the arrow @xmath177 would imply that @xmath172 , which is impossible , because @xmath176 and @xmath181 .
thus @xmath182 , and therefore @xmath183 .
hence @xmath184 .
moreover , there is an arrow @xmath185 . using that @xmath179
, this implies that @xmath186 has an injective successor in @xmath136 and thus @xmath187 .
this establishes our claim that @xmath174 is closed under predecessors in @xmath170 . thus applying the same reasoning as before
, we get that @xmath188 is a local slice in @xmath100 .
now we claim that @xmath189 first , because @xmath190 , we have @xmath191 .
next , @xmath192 contains no injectives , by definition .
thus @xmath193 assume now that @xmath194 belongs to @xmath195
. then @xmath196 and there exists an injective predecessor @xmath197 of @xmath198 in @xmath136 , and since @xmath136 is part of the local slice @xmath135 , there exists a sectional path from @xmath197 to @xmath198 .
applying @xmath199 , we get a sectional path from @xmath200 to @xmath201 in the cluster category .
but this means @xmath202 , which is a contradiction to the hypothesis that @xmath194 . finally , if @xmath203 then @xmath201 is a summand of @xmath20 , which , again , is contradicting the hypothesis that @xmath194 .
following @xcite , let @xmath204 be the full subcategory of @xmath1 consisting of those @xmath205 such that @xmath206 .
[ lem s ] * @xmath204 is hereditary . *
@xmath204 is closed under successors in @xmath1 .
* @xmath1 can be written in the form @xmath207 , \ ] ] where @xmath208 is hereditary , @xmath87 is tilted and @xmath11 is a @xmath87-@xmath208-bimodule .
\(a ) let @xmath209 then @xmath208 is a full subcategory of the hereditary endomorphism algebra of @xmath135 .
therefore @xmath208 is also hereditary , and so @xmath204 is hereditary .
\(b ) let @xmath210 and @xmath211 in @xmath1 .
then there exists a morphism @xmath212 . because @xmath213 is an injective @xmath1-module and
@xmath135 is sincere , there exist a module @xmath214 and a non - zero morphism @xmath215 .
then we have a path @xmath216 , and since @xmath217 , we get that @xmath218 by convexity of the slice @xmath135 in @xmath127 . moreover , since @xmath206 and @xmath136 is closed under predecessors in @xmath135 , it follows that @xmath219 .
thus @xmath220 and this shows ( b ) .
\(c ) this follows from ( a ) and ( b ) .
we recall that the cluster duplicated algebra was introduced in @xcite .
[ cor dup ] the cluster duplicated algebra @xmath221 of @xmath1 is of the form @xmath222\ ] ] where @xmath223 and @xmath224 .
we start by writing @xmath1 in the matrix form of the lemma . by definition ,
@xmath208 consists of those @xmath225 such that the corresponding injective @xmath226 lies in @xmath227 inside the slice @xmath135 .
in particular , the projective dimension of these injectives is at most 1 , hence @xmath228 .
the result now follows upon multiplying by idempotents .
let @xmath13 be a strong sink in @xmath1 .
the reflection at @xmath13 of the algebra @xmath1 is @xmath229\ ] ] where @xmath223 .
[ prop reflection ] the reflection @xmath230 of @xmath1 is a tilted algebra having @xmath167 as a complete slice .
moreover the relation - extensions of @xmath1 and @xmath167 are isomorphic .
we first claim that the support @xmath231 of @xmath167 is contained in @xmath232 .
let @xmath233 .
recall that @xmath234 .
if @xmath203 , then @xmath235 is projective corresponding to a point @xmath236 .
thus @xmath206 and the radical of @xmath237 has no non - zero morphism into @xmath226 .
therefore @xmath238 .
assume next that @xmath239 , that is , @xmath240 , where @xmath241 has an injective successor @xmath213 in @xmath136 . because all sources in @xmath135 are injective , there is an injective @xmath242 and a sectional path @xmath243 .
applying @xmath199 , we obtain a sectional path @xmath244 . in particular the point
@xmath245 belongs to the support of @xmath201 .
assume that there is a point @xmath246 in @xmath208 that is in the support of @xmath201 .
then there exists a nonzero morphism @xmath247 . but
@xmath248 and there is no morphism from @xmath249 to @xmath135 .
therefore @xmath250 .
by the same argument , we show that if @xmath251 , then @xmath250 . finally , all modules of @xmath252 are supported in @xmath87 .
this establishes our claim .
now , by theorem [ thm reflection ] , @xmath253 is a local slice in @xmath254 therefore @xmath255 is a tilted algebra in which @xmath256 is a complete slice .
since the support of @xmath256 is the same as the support of @xmath257 , we are done . we now come to the main result of this section , which states that any two tilted algebras that have the same relation - extension are linked to each other by a sequence of reflections and coreflections .
let @xmath6 be a cluster - tilted algebra and let @xmath25 and @xmath258 be two local slices in @xmath259 .
we write @xmath260 whenever @xmath261 .
[ lem 310 ] let @xmath6 be a cluster - tilted algebra , and @xmath262 be two local slices in @xmath162 .
then there exists a sequence of reflections and coreflections @xmath263 such that @xmath264 given a local slice @xmath25 in @xmath259 such that @xmath25 has injective successors in the transjective component @xmath265 of @xmath62 , let @xmath266 be the rightmost local slice such that @xmath267 . then @xmath266 contains a strong sink @xmath13 , thus reflecting in @xmath13 we obtain a local slice @xmath268 that has fewer injective successors in @xmath265 than @xmath25
. to simplify the notation we define @xmath269 .
similarly , we define @xmath270 , where @xmath271 is the leftmost local slice containing a strong source @xmath205 and @xmath272 .
since we can always reflect in a strong sink , there exist sequences of reflections such that @xmath273 @xmath274 and @xmath275 have no injective successors in @xmath265 .
this implies that @xmath276 .
let @xmath277 thus @xmath278 . [ thm mainreflection ]
let @xmath279 and @xmath280 be two tilted algebras that have the same relation - extension .
then there exists a sequence of reflections and coreflections @xmath263 such that @xmath281 .
let @xmath6 be the common relation - extension of the tilted algebras @xmath279 and @xmath280 .
by @xcite , there exist local slices @xmath282 in @xmath162 such that @xmath283 , for @xmath284 .
now the result follows from lemma [ lem 310 ] and theorem [ thm reflection ] .
let @xmath90 be the path algebra of the quiver @xmath285\ar[ld]\\ & 2\ar[ld ] \\
3\\ & 4\ar[lu]\\ & & 5\ar@<1pt>[lu]\ar@<-1pt>[lu]\ar[ld]\\ & 6 } \ ] ] mutating at the vertices 4,5 , and 2 yields the cluster - tilted algebra @xmath6 with quiver @xmath286\ar@<-2pt>@/_10pt/[lldd]\\ & 2\ar[ur ] \\ 3\ar[ur]\ar@<0pt>@/^15pt/[rrdd]\ar@<-2pt>@/^15pt/[rrdd]\\ & 4\ar@<2pt>[lu]\ar@<-2pt>[lu]\ar[ul]\ar@<1pt>[dd]\ar@<-1pt>[dd]\\ & & 5\ar@<1pt>[lu]\ar@<-1pt>[lu]\\ & 6\ar[ur ] } \ ] ] in the auslander - reiten quiver of @xmath162 we have the following local configuration .
@xmath287 & & \circ & & p(1 ) \\ & 1\ar[ru ] & & 2\ar[rd ] & & { \begin{smallmatrix } 3\\ 5\,5\\4\end{smallmatrix}}\ar[ru ] \\
i(3)\ar[ru]\ar@/^15pt/[rruu ] \ar[rd ] & & \circ & & p(3)\ar[ru]\ar@/^15pt/[rruu]\\ & { \begin{smallmatrix } 5555\\444 \end{smallmatrix}}\ar@<1pt>[rd]\ar@<-1pt>[rd ] & & { \begin{smallmatrix } 55\\4 \end{smallmatrix}}\ar@<1pt>[rd]\ar@<-1pt>[rd ] \ar[ru]\\ & & { \begin{smallmatrix } 555\\44 \end{smallmatrix}}\ar@<1pt>[ru]\ar@<-1pt>[ru ] & & 5 \ar[rd ] \\ &
i(6 ) \ar[ru ] & & \circ & & p(6 ) } \ ] ] where @xmath288 the 6 modules on the left form a rightmost local slice @xmath135 in which both @xmath289 and @xmath290 are sources , so 3 and 6 are strong sinks . for both strong
sinks the subset @xmath152 of the completion consists of the simple module @xmath291 .
the simple module @xmath292 does not lie on a local slice .
the completion @xmath293 is the whole local slice @xmath135 and therefore the reflection @xmath294 is the local slice consisting of the 6 modules on the right containing both @xmath295 and @xmath296 . on the other hand , the completion @xmath297 consists of the four modules @xmath289 , @xmath298 , @xmath299 and @xmath300 , and therefore the reflection @xmath301 is the local slice consisting of the 6 modules on the straight line from @xmath290 to @xmath295 .
this local slice admits the strong sink @xmath302 and the completion @xmath303 in @xmath304 consists of the two modules @xmath290 and @xmath305 .
therefore the reflection @xmath306 is equal to @xmath294 .
thus @xmath307 this example raises the question which indecomposable modules over a cluster - tilted algebra do not lie on a local slice .
we answer this question in a forthcoming publication @xcite .
the objective of this section is to show how to construct those tubes of a tame cluster - tilted algebra which contain projectives .
let @xmath6 be a cluster - tilted algebra of euclidean type , and let @xmath265 be a tube in @xmath62 containing at least one projective .
first , consider the transjective component of @xmath62 .
denote by @xmath308 a local slice in the transjective component that precedes all indecomposable injective @xmath6-modules lying in the transjective component .
then @xmath309 is a tilted algebra having a complete slice in the preinjective component .
define @xmath310 to be a local slice which is a successor of all indecomposable projectives lying in the transjective component .
then @xmath311 is a tilted algebra having a complete slice in the postprojective component .
also , @xmath279 ( respectively , @xmath312 has a tube @xmath313 ( respectively , @xmath314 ) containing the indecomposable projective @xmath279-modules ( respectively , injective @xmath280-modules ) corresponding to the projective @xmath6-modules in @xmath265 ( respectively , injective @xmath6-modules in @xmath265 ) . an indecomposable projective @xmath14 ( respectively , injective @xmath15 )
@xmath6-module that lies in a tube , is said to be a _ root projective _
( respectively , a _ root injective _ ) if there exists an arrow in @xmath6 between @xmath13 and @xmath205 , where the corresponding indecomposable projective @xmath237 lies in the transjective component of @xmath62 .
let @xmath315 be the coray in @xmath313 passing through the projective @xmath279-module that corresponds to the root projective @xmath316 in @xmath265 .
similarly , let @xmath317 be the ray in @xmath314 passing through the injective that corresponds to the root injective @xmath318 in @xmath265 . recall that if @xmath90 is hereditary and @xmath319 is a tilting module
, then there exists an associated torsion pair @xmath320 in @xmath321 , where @xmath322 [ lem tube1 ] with the above notation * @xmath323 is a coray in @xmath265 passing through @xmath316 .
* @xmath324 is a ray in @xmath265 passing through @xmath318 .
since @xmath279 is tilted , we have @xmath325 where @xmath20 is a tilting module over a hereditary algebra @xmath90 . as seen in the proof of theorem 5.1 in @xcite , we have a commutative diagram @xmath326\ar[r]^{\text{hom}_a(t,-)}&\mathcal{y}(t)\ar[d]^{-\otimes_{c_1 } b}\\ \mathcal{c}_a\ar[r]^{\text{hom}_{\mathcal{c}_a}(t,-)}&\text{mod}\,b\;\;}\ ] ] where @xmath327 .
let @xmath328 be the tube in @xmath321 corresponding to the tube @xmath265 in @xmath259 .
by what has been seen above , we have a commutative diagram @xmath329^{\text{hom}_a(t,-)}\ar[dr]_-{\text{hom}_{\mathcal{c}_a}(t,-)\;\;\;\;\;}&\mathcal{t}_1\ar[d]^{-\otimes_{c_1 } b}\\ & \mathcal{t}_1\otimes_{c_1 } b \subset \mathcal{t}\;\;.}\ ] ] let @xmath330 be any coray in @xmath313 , so it can be lifted to a coray @xmath331 in @xmath332 via the functor @xmath333 .
if we apply @xmath334 to this lift , we obtain a coray in @xmath335 .
thus , any coray in @xmath313 induces a coray in @xmath265 .
let @xmath315 be the coray passing through the root projective @xmath336 .
then @xmath337 is the coray passing through @xmath338 .
this proves ( a ) and part ( b ) is proved dually .
however , we must still justify that the ray @xmath337 and the coray @xmath339 actually intersect ( and thus lie in the same tube of @xmath62 ) . because @xmath340 , we have @xmath341 , and @xmath316 lies in a tube @xmath265 .
it is well - known that the injective @xmath318 also lies in @xmath265 . in particular
, we have the following local configuration in @xmath265 , where @xmath342 is an indecomposable summand of the radical of @xmath316 and @xmath343 an indecomposable summand of the quotient of @xmath318 by its socle .
@xmath344 & & \circ \ar@{-->}[dr ] & & p_b(i)\\ & j\ar[dr]\ar@{-->}[ur]&&r\ar[ur]\\ & & n\ar[ur]}\ ] ] now @xmath345 is coinduced , and we have shown above that the ray containing it is also coinduced . because @xmath346 , this is the ray @xmath347 . therefore , this ray and this coray lie in the same tube , so must intersect in a module @xmath348 , where there exists an almost split sequence @xmath349&j\ar[r ] & n \ar[r ] & r \ar[r ] & 0.}\ ] ] knowing the ray @xmath347 and the coray @xmath337 for every root projective @xmath316 in @xmath265
, one may apply the knitting procedure to construct the whole of @xmath265 . in this way
, @xmath265 can be determined completely .
next we show that all modules over a tilted algebra lying on the same coray change in the same way under the induction functor .
[ lem tube2 ] let @xmath90 be a hereditary algebra of euclidean type , @xmath20 a tilting @xmath90-module without preinjective summands and let @xmath350 be the corresponding tilted algebra .
let @xmath328 be a tube in @xmath351 and @xmath352 an indecomposable summand of @xmath20 , such that @xmath353 .
then there exists an @xmath90-module @xmath11 on the mouth of @xmath328 such that we have @xmath354 in @xmath127 .
in particular , the module @xmath355 lies on the mouth of the tube @xmath356 in @xmath127 .
the injective @xmath1-module @xmath357 is given by @xmath358 where the first identity holds by ( * ? ? ?
* proposition vi 5.8 ) and the second identity is the auslander - reiten formula .
moreover , since @xmath359 lies in the tube @xmath328 and @xmath20 has no preinjective summands , we have @xmath360 only if @xmath200 lies in the hammock starting at @xmath359 .
furthermore , if @xmath200 is a summand of @xmath20 then it must lie on a sectional path starting from @xmath359 because @xmath361 .
this shows that a point @xmath362 is in the support of @xmath357 if and only if there is a sectional path @xmath363 in @xmath328 .
we shall distinguish two cases .
if @xmath359 lies on the mouth of @xmath328 then let @xmath364 be the ray starting at @xmath359 and denote by @xmath365 the last summand of @xmath20 on this ray .
let @xmath366 be the direct predecessor of @xmath365 not on the ray @xmath364 .
thus we have the following local configuration in @xmath328 .
@xmath367&&\tau t_i \ar[dr ] & & t_i \ar[dr]\\ & \ar[ur]\ar[dr]&&\ar[ur]\ar[dr]&&\ar@{ ..
}[dr]\\ & & \ar[ur]\ar@{ .. }[dr]&&\ar@{
.. }[dr]&&\ar[dr]\\ & & & \ar[dr]&&\tau t_1\ar[dr]\ar[ur]&&t_1\ar[dr]&\\ & & & & \tau l_1\ar[ur]\ar[dr]&&l_1\ar[ur]&&\tau^{-1 } l_1 \\ & & & & & e_1\ar[ur]}\ ] ] then @xmath357 is uniserial with simple top @xmath298 .
moreover there is a short exact sequence @xmath349&\tau t_i\ar[r ] & l_1\ar[r ] & t_1\ar[r ] & 0}\ ] ] and applying @xmath368 yields @xmath369&\text{hom}_a(t , l_1)\ar[r ] & p_c(1 ) \ar[r]^{f } & i_c(i ) \ar[r ] & \text{ext}^1(t , l_1)\ar[r]&0}\ ] ] by the auslander - reiten formula , we have @xmath370 and this is zero because @xmath365 is the last summand of @xmath20 on the ray @xmath364 .
thus the sequence ( [ s1 ] ) is short exact , the morphism @xmath371 is a projective cover , because @xmath357 is uniserial , and hence @xmath372 applying @xmath373 yields @xmath374 let @xmath375 be the indecomposable direct predecessor of @xmath366 such that the almost split sequence ending at @xmath366 is of the form @xmath376&\tau l_1 \ar[r ] & e_1\oplus \tau t_1 \ar[r ] & l_1 \ar[r ] & 0}\ ] ] we claim that @xmath377 .
recall that @xmath366 is not a summand of @xmath20 because @xmath378 is non projective . also , recall that @xmath365 is the last summand of @xmath20 on the ray @xmath364 .
suppose @xmath379 , thus @xmath380 .
then it follows that there is a summand of @xmath20 on the ray @xmath381 that is a successor of @xmath382 .
let @xmath383 denote the first such indecomposable summand .
@xmath367&&\tau t_1\ar[dr ] & & t_1\ar[dr]\\ & \tau l_1\ar[ur]\ar[dr ] & & l_1 \ar[dr]\ar[ur]&&\tau^{-1}l_1\ar[dr]\\ & & e_1 \ar[ur ] & & \tau^{-1 } e_1\ar[dr ] \ar[ur ] & & \ar@{ ..
}[dr]\\ & & & & & \ar@{ .. }[dr ] & & n \ar[dr]\\ & & & & & & t^1\ar[ur]\ar[dr ] & & \ar@{ ..
}[dr]\\ & & & & & & & \ar@{ .. }[dr ] & & \omega \\ & & & & & & & & \tau\omega & & & \\ } \ ] ] then we have a short exact sequence @xmath349 & l_1\ar[r]^-{h}&t_1\oplus t^1 \ar[r ] & n \ar[r ] & 0}\ ] ] with @xmath246 an @xmath384-approximation . applying @xmath385 yields @xmath386&\text{hom}_a(n , t)\ar[r ] & \text{hom}_a(t_1\oplus t^1 , t)\ar[r]^-{h^ * } & \text{hom}_a(l_1 , t)\\ & \hspace{2cm}\ar[r]&\text{ext}^1_a(n , t)\ar[r]&0\hspace{1.5cm}}\ ] ] and since @xmath246 is an @xmath384-approximation , the morphism @xmath387 is surjective .
thus @xmath388 .
on the other hand , @xmath389 generates @xmath348 , so @xmath390 , and thus @xmath391 .
but then both @xmath392 and we see that @xmath348 is a summand of @xmath20 . this is a contradiction to the assumption that @xmath365 is the last summand of @xmath20 on the ray @xmath364 .
thus @xmath393 . therefore , in the almost split sequence ( [ s2 ] ) , we have @xmath394 and @xmath395 . moreover , all predecessors of @xmath396 on the ray @xmath397 are also in @xmath398 because the morphisms on the ray are injective .
since @xmath399 is an equivalence of categories , it follows that @xmath400 has only one direct predecessor @xmath401 in @xmath8 and this irreducible morphism is surjective .
the kernel of this morphism is @xmath402 where @xmath403 is the torsion radical .
thus we get @xmath404 we will show that @xmath405 lies on the mouth of @xmath328 and this will complete the proof in case 1 .
let @xmath11 be the indecomposable @xmath90-module on the mouth of @xmath328 such that the ray starting at @xmath11 passes through @xmath406 .
thus @xmath11 is the starting point of the ray @xmath407 .
then there is a short exact sequence of the form @xmath408 & m \ar[r ] & \tau_a l_1 \ar[r ] & \tau_a t_1 \ar[r ] & 0}\ ] ] with @xmath409 .
we claim that @xmath410 .
suppose to the contrary that @xmath411 .
since @xmath412 lies on the mouth of @xmath328 , this implies that there is a direct summand @xmath383 of @xmath20 which lies on the ray @xmath397 starting at @xmath412 .
since @xmath20 is tilting , @xmath383 can not be a predecessor of @xmath396 on this ray and since @xmath366 is not a summand of @xmath20 , we also have @xmath413 .
thus @xmath383 is a successor of @xmath366 on the ray @xmath381 .
this is impossible since such a @xmath383 would satisfy @xmath414 contradicting the fact that @xmath377 .
therefore , @xmath415 and the sequence ( [ s3 ] ) is the canonical sequence for @xmath406 in the torsion pair @xmath320 .
this shows that @xmath416 and hence @xmath417 as desired .
now suppose that @xmath359 does not lie on the mouth of @xmath328 .
let @xmath418 denote the ray passing through @xmath359 and @xmath419 the coray passing through @xmath359 .
denote by @xmath365 the last summand of @xmath20 on @xmath418 , by @xmath420 the last summand of @xmath20 on @xmath419 , and by @xmath421 the direct predecessor of @xmath200 which does not lie on @xmath422 .
note that @xmath423 does not exist if @xmath420 lies on the mouth of @xmath328 , and in this case we let @xmath424 .
thus we have the following local configuration in @xmath328 .
@xmath425 & & & & & & & & & & l_2 \ar[dr ] & & \tau^{-1}l_2\\ & \ar@{ .. }[dr]&&&&&&&&\tau t_2\ar[ur]\ar[dr]\ar@{ .. }[dl ] & & t_2\ar[ur]&&\\ & & \ar@{ .. }[dr]&&&&&&&&\ar[ur]\ar@{ .. }[dl]\\ & & & \ar@{ .. }[dr]&&\ar[dr]\ar@{ .. }[ur]&&\ar[dr]\ar@{ .. }[ur]&&&&&&&&\\ & & & & \tau^2 t_i\ar[ur]\ar[dr ] & & \tau t_i \ar[dr]\ar[ur ] & & t_i \ar[ur]\ar[dr]\\ & & & & & \ar[ur]\ar@{ .. }[dr]&&\ar[ur]\ar@{ .. }[dr]&&\ar@{ .. }[dr]&&\\ & & & & & & \ar[dr]&&\tau t_1\ar[dr ] & & t_1\ar[dr ] \\ & & & & & & & \tau l_1 \ar[ur ] & & l_1\ar[ur ] & & \tau^{-1}l_1}\ ] ] the injective @xmath1-module @xmath426 is biserial with top @xmath427
. moreover , there is a short exact sequence
@xmath349&\tau t_i \ar[r ] & l_1\oplus
l_2 \oplus t_i \ar[r ] & t_1\oplus t_2 \ar[r ] & 0.}\ ] ] applying @xmath333 yields the following exact sequence . @xmath428&\text{hom}_a(t , l_1\oplus l_2)\oplus p_c(i ) \ar[r ] & p_c(1)\oplus p_c(2 ) \ar[r]^-{f}&i_c(i)\\ & \hspace{3cm}\ar[r]&\text{ext}^1_a(t , l_1\oplus l_2)\ar[r]&0.}\ ] ] by the same argument as in case 1 , using that @xmath365 and @xmath420 are the last summands of @xmath20 on @xmath418 and @xmath419 respectively , we see that @xmath429 .
therefore , the sequence ( [ s4 ] ) is short exact .
moreover , the morphism @xmath371 is a projective cover and thus @xmath430 applying @xmath373 yields @xmath431 by the same argument as in case 1 we see that @xmath432 where @xmath11 is the indecomposable @xmath90-module on the mouth of @xmath328 such that the ray starting at @xmath11 passes through @xmath433 . in other words , @xmath11 is the starting point of the ray @xmath434 .
therefore , it only remains to show that @xmath435 .
to do so , it suffices to show that @xmath423 is a summand of @xmath20 .
we have already seen that @xmath436 .
we show now that we also have @xmath437 .
suppose the contrary .
then there exists a non - zero morphism @xmath438 . composing it with the irreducible injective morphism @xmath439 yields a non - zero morphism in @xmath440 .
but this is impossible since @xmath20 is tilting
. thus we have @xmath441 and thus @xmath423 is a summand of @xmath20 , the module @xmath442 is projective and @xmath443 .
this completes the proof .
the module @xmath11 in the statement of the lemma is the starting point of the ray passing through @xmath444 .
[ cor5 ] let @xmath445 be as in lemma [ lem tube2 ] , and let @xmath446 , with @xmath447 .
let @xmath448 be two modules lying on the same coray in the tube @xmath356 in @xmath127 .
then @xmath449 and thus the two projections @xmath450 and @xmath451 have isomorphic kernels . for all @xmath1-modules
@xmath201 we have @xmath452 where the first isomorphism is ( * ? ? ?
* proposition 3.3 ) and the second is ( * ? ? ?
* proposition 4.1 ) .
since @xmath20 has no preinjective summands , and @xmath201 is regular , the only summand of @xmath453 for which @xmath454 can be nonzero , must lie in the same tube as @xmath201 . by the lemma ,
the only summands of @xmath453 in the tube lie on the mouth of the tube .
let @xmath11 denote an indecomposable @xmath1-module on the mouth of a tube .
then @xmath455 we summarize the results of this section in the following proposition .
[ prop5 ] * let @xmath456 be the coray in @xmath457 passing through the projective @xmath279-module corresponding to the root projective @xmath316 then @xmath458 is a coray in @xmath459 passing through @xmath316 .
furthermore all modules in @xmath460 are extensions of modules of @xmath456 by the same module @xmath461 .
* let @xmath462 be the ray in @xmath463 passing through the injective @xmath280-module corresponding to the root injective @xmath318 then @xmath464 is a ray in @xmath459 passing through @xmath318 .
furthermore all modules in @xmath464 are extensions of modules of @xmath462 by the same module @xmath465 .
\(a ) the first statement is lemma [ lem tube1 ] , and the second statement is a restatement of corollary [ cor5 ] .
let @xmath6 be the cluster - tilted algebra given by the quiver @xmath466^{\lambda}\ar@<-2pt>[rr]_{\beta}&&5\ar[dl]^{\epsilon}\\ & 3\ar[ul]^{\alpha}\ar[dr]^{\delta}\\ 2\ar[ur]^{\gamma}&&4\ar[ll]^{\sigma}}\ ] ] bound by @xmath467 .
the algebras @xmath279 and @xmath280 are respectively given by the quivers @xmath466^{\lambda}\ar@<-2pt>[rr]_{\beta}&&5 & & & & 1\ar@<2pt>[rr]^{\lambda}\ar@<-2pt>[rr]_{\beta}&&5\ar[dl]^{\epsilon } \\ & 3\ar[dr]^{\delta } \ar[ul]^{\alpha}&&&\text{and } & & & 3\ar[dr]^{\delta}\\ 2\ar[ur]^{\gamma}&&4 & & & & 2\ar[ur]^{\gamma}&&4}\ ] ] with the inherited relations .
we can see the tube in @xmath468 below and the coray passing through the root projective @xmath469 is given by @xmath470 & { \begin{smallmatrix}1\\5\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}3\\4\;1\\\;\;\;5\end{smallmatrix}}\ar[r ] & { \begin{smallmatrix}3\\1\\5\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}2\\3\\1\\5\end{smallmatrix}}.}\ ] ] @xmath471\ar@{--}[dd]\\ & & & & & & & { \begin{smallmatrix}3\\4\;1\\\;\;\;5\end{smallmatrix}}\\ & & & & { \begin{smallmatrix}2\\3\\1\\5\end{smallmatrix}}\ar[dr]&&{\begin{smallmatrix}1\\5\end{smallmatrix}}\ar[ur]\ar@{--}[d]\\ & { \begin{smallmatrix}4\end{smallmatrix}}\ar[dr]\ar@{--}[dd]&&{\begin{smallmatrix}3\\1\\5\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}2\;\;\\3\;\;\\1\;1\\\;5\;5\end{smallmatrix}}\ar[ur]&&\\ & & { \begin{smallmatrix}3\\4\;1\\\;\;\;5\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}3\\1\;1\\\;5\;5\end{smallmatrix}}\ar[ur]&&\\ & { \begin{smallmatrix}1\\5\end{smallmatrix}}\ar[ur]\ar[dr]\ar@{--}[d]&&{\begin{smallmatrix}3\\4\;1\;1\\\;\;\;\;5\;5\end{smallmatrix}}\ar[ur]\\ \ar[ur]&&{\begin{smallmatrix}1\;1\\\;5\;5\end{smallmatrix}}\ar[ur]}\ ] ] dually , the ray in @xmath472 passing through the root injective @xmath473 is given by @xmath474&{\begin{smallmatrix}1\\5\\3\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}1\;\;\;\\5\;2\\3\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}1\\5\end{smallmatrix}}\ar[r]&\dots } \ ] ] the root projective @xmath475 lies on the coray @xmath476 & { \begin{smallmatrix}1\\5\\3\\4\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}3\\4\;1\\\;\;\;5\\\;\;\;3\\\;\;\;4\end{smallmatrix}}\ar[r ] & { \begin{smallmatrix}3\\1\\5\\3\\4\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}2\\3\\1\\5\\3\\4\end{smallmatrix}}}\ ] ] and the root injective @xmath477 lies on the ray @xmath478&{\begin{smallmatrix}2\\3\\1\\5\\3\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}2\;\;\;\\3\;\;\;\\1\;\;\;\\5\;2\\3\end{smallmatrix}}\ar[r]&{\begin{smallmatrix}2\\3\\1\\5\end{smallmatrix}}\ar[r]&\dots } \ ] ] note that by proposition [ prop5 ] , every module in @xmath479 is an extension of a module in @xmath315 by @xmath480 .
similarly , every module in @xmath324 is an extension of a module in @xmath317 by @xmath481 . applying the knitting algorithm we obtain the tube in @xmath482 containing both @xmath337 and @xmath324 .
@xmath483\ar@{--}[dd]&&\circ&&{\begin{smallmatrix}2\\3\\1\\5\\3\\4\end{smallmatrix}}\ar[dr]&&\circ&&{\begin{smallmatrix}4\\2\end{smallmatrix}}\ar@{--}[dd]\\ & { \begin{smallmatrix}4\end{smallmatrix}}\ar[dr]&&{\begin{smallmatrix}3\\1\\5\\3\\4\end{smallmatrix}}\ar[dr]\ar[ur]&&{\begin{smallmatrix}2\\3\\1\\5\\3\end{smallmatrix}}\ar[dr]&&{\begin{smallmatrix}2\end{smallmatrix}}\ar[ur]\\ \circ \ar@{--}[dd]&&{\begin{smallmatrix}3\\4\;1\\\;\;\;5\\\;\;\;3\\\;\;\;4\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}3\\1\\5\\3\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}2\;\;\;\\3\;\;\;\\1\;\;\;\\5\;2\\3\end{smallmatrix}}\ar[ur]\ar[dr]&&\circ\ar@{--}[dd]\\ & { \begin{smallmatrix}1\\5\\3\\4\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}3\\4\;1\\\;\;\;\;5\\\;\;\;3\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}3\;\;\;\\1\;\;\;\\5\;2\\3\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}2\\3\\1\\5\end{smallmatrix}}\ar[dr]\\ { \begin{smallmatrix}2\;\;\;\\3\;\;\;\\\;1\;1\\\;\;5\;5\\\;\;\;\;\;3\\\;\;\;\;\;4\end{smallmatrix}}\ar[ur]\ar[dr]\ar@{--}[d]&&{\begin{smallmatrix}1\\5\\3\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}3\;\;\;\\4\;1\\\;\;\;5\;2\\\;\;\;\;\;\;3\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}3\\1\\5\end{smallmatrix}}\ar[ur]\ar[dr]&&{\begin{smallmatrix}2\;\;\;\\3\;\;\;\\\;1\;1\\\;\;5\;5\\\;\;\;\;\;3\\\;\;\;\;\;4\end{smallmatrix}}\ar@{--}[d]\\ & \ar[ur]&&\ar[ur]&&\ar[ur]&&\ar[ur]&}\ ] ]
let @xmath6 be cluster - tilted of euclidean type @xmath19 and let @xmath484 .
then there exists @xmath485 tilting such that @xmath486 . because @xmath19 is euclidean
, @xmath95 contains at most 3 exceptional tubes .
denote by @xmath487 the direct sums of those summands of @xmath20 that respectively lie in the transjective component and in the three exceptional tubes .
let @xmath89 be a hereditary category that is derived equivalent to @xmath351 and such that @xmath89 is not the module category of a hereditary algebra .
then @xmath89 is of the form @xmath489 , where @xmath490 consist of tubes , and @xmath100 is a transjective component , see @xcite .
let @xmath491 , @xmath492 be the direct sum of all indecomposable summands of @xmath20 lying in @xmath493 , @xmath494 respectively .
we define two subspaces @xmath495 and @xmath342 of @xmath6 as follows .
@xmath496 the transjective component of @xmath162 contains a left section @xmath497 and a right section @xmath498 , see @xcite .
thus @xmath499 are local slices , @xmath497 has no projective predecessors , and @xmath500 has no projective successors in the transjective component .
define @xmath111 to be the two - sided ideal of @xmath6 generated by @xmath501 and the two subspaces @xmath495 and @xmath342 .
thus @xmath502 we call @xmath111 the _ partition ideal _ induced by the partition @xmath503 .
we have @xmath507 , where the last equality is as @xmath2-vector spaces . using the decomposition @xmath508 , we see that @xmath6 is equal to @xmath509 where all hom spaces are taken in @xmath488 . on the other hand
, @xmath510 is a quasi - tilted algebra .
thus in order to prove that @xmath1 is quasi - tilted it suffices to show that @xmath111 is the ideal generated by @xmath511 but this follows from the definition of @xmath495 and @xmath342 and the fact that the annihilators of the local slices @xmath497 and @xmath498 are given by the morphisms in @xmath512 that factor through the lifts of the corresponding local slice in the cluster category .
more precisely , @xmath513 and thus @xmath514 where we used the fact that @xmath515 .
this completes the proof that @xmath1 is quasi - tilted .
now assume that @xmath506 .
then @xmath517 and thus @xmath111 is generated by @xmath518 , and this is equal to @xmath519 on the other hand , @xmath517 implies that @xmath520 and since @xmath521 , this implies that @xmath522 is the annihilator of a local slice .
therefore @xmath504 is tilted by @xcite .
the case where @xmath505 is proved in a similar way . conversely , assume @xmath1 is tilted .
then @xmath523 for some local slice @xmath304 in @xmath162 .
we show that @xmath522 or @xmath524 .
suppose to the contrary that @xmath304 has both a predecessor and a successor in @xmath525 .
then there exists an arrow @xmath526 in the quiver of @xmath6 such that @xmath527 and @xmath528 .
but by definition of @xmath529 and @xmath342 , we see that this is impossible .
[ thm ctaqt2 ] if @xmath1 is quasi - tilted of euclidean type and @xmath7 then @xmath530 where @xmath531 is a right section in the postprojective component of @xmath1 and @xmath532 is a left section in the preinjective component .
@xmath1 being quasi - tilted implies that there is a hereditary category @xmath89 with a tilting object @xmath20 such that @xmath101 .
moreover , @xmath533 is the corresponding cluster - tilted algebra . as before we use the decomposition @xmath534 .
then the algebras @xmath535 are tilted .
let @xmath536 and @xmath537 be complete slices in @xmath538 and @xmath539 respectively .
note that @xmath536 lies in the postprojective component and @xmath532 lies in the preinjective component of their respective module categories .
then @xmath1 is a branch extension of @xmath120 by the module @xmath540 similarly @xmath1 is a branch coextension of @xmath122 by the module @xmath541 observe that the postprojective component of @xmath120 does not change under the branch extension , and the preinjective component of @xmath122 does not change under the branch coextension .
therefore @xmath536 is a right section in the postprojective component of @xmath1 and @xmath532 is a left section in the preinjective component .
moreover , by construction , we have @xmath542 and therefore @xmath543 this completes the proof .
assume first that @xmath1 is tilted .
then , because of @xcite , there exists a local slice @xmath135 in the transjective component of @xmath459 such that @xmath545 .
otherwise , assume that @xmath1 is quasi - tilted but not tilted .
then , because of @xcite , there exists a hereditary category @xmath89 of the form @xmath546 and a tilting object @xmath20 in @xmath89 such that @xmath101 . because of theorem [ thm ctaqt ] we get @xmath504 where @xmath111 is the partition ideal induced by the given partition of @xmath89 .
let @xmath6 be the cluster - tilted algebra of type @xmath547 given by the quiver @xmath548 & & 7\ar[ll]_{\epsilon}\\ & 6\ar[ru]\ar[rdd]^(0.33){{{\beta}_3 } } \\ & 5\ar[rd]^(0.4){{\beta}_2\quad}\\ 1\ar[ruu]^(0.65 ) { { \alpha}_3}\ar[ru]^(0.6 ) { { \alpha}_2}\ar[rd]_(0.6){{\alpha}_1 } & & 2\ar@<1.5pt>[ll]\ar@<-1.5pt>[ll]\\ & 3\ar[ru]_(0.4){{\beta}_1}\ar[rd ] \\ & & 4 } \ ] ] as usual let @xmath359 denote the indecomposable summand of @xmath20 corresponding to the vertex @xmath30 of the quiver .
in this example @xmath20 has two transjective summands @xmath549 , and the other summands lie in three different tubes .
@xmath550 lie in a tube @xmath551 , @xmath552 lies in a tube @xmath553 and @xmath554 lie in a tube @xmath555 . choosing a partition ideal corresponds to choosing a subset of tubes to be predecessors of the transjective component .
thus there are 8 different partition ideals corresponding to the 8 subsets of @xmath556 .
if the tube @xmath557 is chosen to be a predecessor of the transjective component , then the arrow @xmath558 is in the partition ideal . and if @xmath559 is not chosen to be a predecessor of the transjective component , then it is a successor and consequently the arrow @xmath560 is in the partition ideal .
the arrow @xmath561 is always in the partition ideal since it corresponds to a morphim from @xmath562 to @xmath563 in the derived category .
the quiver of the corresponding quasi - tilted algebra @xmath565 is obtained by removing the generating arrows from the quiver of @xmath6 .
exactly 2 of these 8 algebras are tilted , and these correspond to cutting @xmath566 , respectively @xmath567 .
c. amiot , cluster categories for algebras of global dimension 2 and quivers with potential , _ ann .
fourier _ * 59 * no 6 , ( 2009 ) , 25252590 . i. assem , left sections and the left part of an artin algebra , _ colloq .
_ * 116 * ( 2009 ) , no .
2 , 273300 . i. assem , t. brstle , g. charbonneau - jodoin and p. g. plamondon , gentle algebras arising from surface triangulations
. _ algebra number theory _ * 4 * ( 2010 ) , no .
2 , 201229 . , cluster - tilted algebras as trivial extensions , _ bull . lond .
_ * 40 * ( 2008 ) , 151162 . , cluster - tilted algebras and slices , _ j. of algebra _ * 319 * ( 2008 ) , 34643479 .
, on the galois covering of a cluster - tilted algebra , _ j. pure appl .
alg . _ * 213 * ( 7 ) ( 2009 ) 14501463 . , cluster - tilted algebras without clusters , _ j. algebra _ * 324 * , ( 2010 ) , 24752502 .
i. assem , r. schiffler and k. serhiyenko , modules that do not lie on local slices , in preparation .
i. assem , d. simson and a. skowroski , _ elements of the representation theory of associative algebras , 1 : techniques of representation theory _ , london mathematical society student texts 65 , cambridge university press , 2006 .
m. auslander , i. reiten and s.o .
smal , _ representation theory of artin algebras _ cambridge studies in advanced math .
36 , ( cambridge university press , cambridge , 1995 ) .
m. barot and s. trepode , cluster tilted algebras with a cyclically oriented quiver .
_ comm . algebra_*41 * ( 2013 ) , no .
10 , 36133628 . , from iterated tilted to cluster - tilted algebras , _ adv .
_ * 223 * ( 2010 ) , no .
4 , 14681494 .
m. a. bertani - kland , s. oppermann and a wrlsen , constructing tilted algebras from cluster - tilted algebras , _
j. algebra _ * 323 * ( 2010 ) , no . 9 , 24082428 .
a. b. buan , r. marsh , m. reineke , i. reiten and g. todorov , _ tilting theory and cluster combinatorics _ , adv . math . * 204 * ( 2006 ) , no . 2 , 572 - 618 . , cluster - tilted algebras , _ trans .
_ * 359 * ( 2007 ) , no .
1 , 323332 ( electronic ) . e. fernndez , n. i. pratti and s. trepode , on @xmath569-cluster tilted algebras and trivial extensions , _ j. algebra _ * 393 * ( 2013 ) , 132141 .
s. fomin and a. zelevinsky , cluster algebras i : foundations , _ j. amer . math .
* 15 ( 2002 ) , 497529 .
d. happel , a characterization of hereditary categories with tilting object .
_ * 144 * ( 2001 ) , no .
2 , 381398 .
d. happel , i. reiten and s. smal , tilting in abelian categories and quasitilted algebras .
* 120 * ( 1996 ) , no .
h. lenzing and a. skowroski , quasi - tilted algebras of canonical type , _ colloq . math . _
* 71 * ( 1996 ) , no .
2 , 161181 . b. keller and i. reiten , cluster - tilted algebras are gorenstein and stably calabi - yau , _ adv .
_ * 211 * ( 2007 ) , no .
1 , 123151 . m. oryu and r. schiffler , on one - point extensions of cluster - tilted algebras , em j. algebra * 357 * ( 2012 ) , 168182 .
i. reiten , cluster categories .
proceedings of the international congress of mathematicians .
volume i , 558594 , hindustan book agency , new delhi , 2010 .
16 - 02 c.m .
ringel , the regular components of the auslander - reiten quiver of a tilted algebra .
_ chinese ann .
b _ * 9 * ( 1988 ) , no .
1 , 118 . c. m. ringel , representation theory of finite - dimensional algebras .
representations of algebras ( durham , 1985 ) , 779 , london math .
lecture note ser .
, 116 , cambridge univ . press , cambridge , 1986 .
r. schiffler , _ quiver representations _ , cms books in mathematics , springer international publishing , 2014 .
r. schiffler and k. serhiyenko , induced and coinduced modules over cluster - tilted algebras , preprint , arxiv:1410.1732 .
r. schiffler and k. serhiyenko , injective presentations of induced modules over cluster - tilted algebras , preprint , arxiv:1604.06907 a. skowroski , tame quasi - tilted algebras , _ j. algebra _ * 203 * ( 1998 ) , no .
2 , 470490 . * | in this paper , we prove that relation - extensions of quasi - tilted algebras are 2-calabi - yau tilted . with the objective of describing the module category of a cluster - tilted algebra of euclidean type ,
we define the notion of reflection so that any two local slices can be reached one from the other by a sequence of reflections and coreflections .
we then give an algorithmic procedure for constructing the tubes of a cluster - tilted algebra of euclidean type .
our main result characterizes quasi - tilted algebras whose relation - extensions are cluster - tilted of euclidean type . |
a common feature of spatially extended systems that are driven far - from - equilibrium is spatiotemporal chaos where the dynamics are aperiodic in space and time @xcite .
important examples include the dynamics of the atmosphere and climate @xcite ; fluid convection @xcite ; the convection of biological organisms in the oceans @xcite , the transition to chaos in excitable media @xcite ; and the complicated dynamics of systems of reacting , advecting and diffusing chemicals @xcite . despite intense theoretical and experimental investigation
many important challenges remain regarding how to characterize , control , and predict the dynamics of these large complex systems .
a unifying feature of many of these systems is that the dimension of the attractors describing their dynamics in phase space is very large .
this severely limits the use of sophisticated techniques of chaotic time - series analysis @xcite and of powerful geometry based descriptions of dynamical systems @xcite .
however , the use of lyapunov exponents to compute the fractal dimension does not suffer from these limitations and provides an important window into fundamental aspects of the dynamics of high dimensional chaotic systems .
furthermore , the spectrum of lyapunov exponents are inaccessible to experimental measurement using currently available techniques which leaves their numerical computation as an important way to gain new insights .
a physical understanding of the dimension and how it varies with system parameters can provide fundamental knowledge regarding the underlying nature of spatiotemporal chaos that can be used to guide the development of improved theoretical descriptions for experimentally accessible systems .
a defining feature of chaos is the sensitive dependence of the system dynamics on initial conditions and the exponential separation of nearby trajectories in phase space @xcite .
the rate of separation is quantified by the spectrum of lyapunov exponents @xmath0 where @xmath1 with @xmath0 arranged in descending order and @xmath2 is the total number of degrees of freedom in the system . for partial differential equations , although the phase space is infinite , the value of @xmath2 is expected to be very large but finite corresponding to the dimension of the inertial manifold ( c.f .
@xcite ) . for a system of coupled ordinary differential equations , as we study here , the value of @xmath2 is equal to the number of independent variables .
the summation of the first @xmath3 exponents ( where @xmath4 ) describes the growth of an @xmath3-dimensional ball of initial conditions in @xmath2-dimensional phase space .
the exact interpolated number of exponents required for the sum to vanish is the dimension of the ball that will neither grow nor shrink and is an estimate of the fractal dimension @xmath5 of the strange attractor . using a linear interpolation yields the well known kaplan - yorke formula @xmath6 where @xmath7 is the largest integer such that the sum of the first @xmath7 lyapunov exponents is non - negative @xcite .
the magnitude of @xmath5 is an approximate value of the number of degrees of freedom acting in the system @xcite .
ruelle @xcite initially conjectured , and numerical simulation later confirmed ( c.f .
@xcite ) , that the fractal dimension is extensive for large chaotic systems .
more precisely , @xmath5 increases linearly with the volume of the system @xmath8 where @xmath9 is a characteristic length and @xmath10 is the number of spatially extended dimensions .
the extensivity of @xmath5 is a consequence of spatial disorder .
if two subsystems are sufficiently far apart their coupling is weak and they contribute independently and additively to the overall dimension @xcite .
cross and hohenberg @xcite used this to define a natural chaotic length scale as @xmath11 where a volume @xmath12 will contain , on average , one degree of freedom . using the above arguments
, one intriguing possibility is that underlying spatiotemporal chaos are structures of volume @xmath13 .
this raises an interesting question : how does the dimension vary for changes in system volume that are smaller than @xmath12 ?
one possibility is that @xmath5 is proportional to the system size only on average . in this case
, the variation of @xmath5 would have a stepwise structure with system size where the steps correspond to the addition of new degrees of freedom as the system grows large enough to accommodate them .
a smoother transition may also be possible where the new degrees of freedom are able to stretch or compress to match the system size . in light of this , the variation of @xmath5 for small changes in system size can shed important physical insight upon the basic nature and composition of spatiotemporal chaos ( c.f .
@xcite ) .
questions relating to the precise variation of the fractal dimension with system size have been explored numerically for several important model equations .
ohern _ et al .
_ explored the extensive chaos of a 2d coupled - map lattice with an ising - like transition @xcite .
the fractal dimension was found to exhibit significant deviations from extensivity near the onset of spatiotemporal chaos which rapidly decayed with increasing system size ( here the number of lattice sites ) .
xi _ et al . _
@xcite explored the 1d nikolaevskii equation and tajima and greenside @xcite explored the 1d kuramoto - sivashinsky equation . in both of these studies
extensive chaos was found and a systematic exploration of the dimension was performed in the extensive regime .
the fractal dimension remained linear to within the precision of the calculations to yield what is now referred to as microextensive chaos .
it is important to note that neither study focussed upon the approach to extensive chaos where deviations may be more accessible . at the larger system sizes
explored it remains a possibility that deviations in the fractal dimension were too small to detect .
fishman and egolf explored the 1d complex ginzburg - landau equation over a range of system sizes including the approach to extensive chaos @xcite .
significant deviations from extensivity were found to occur on a length scale consistent with @xmath14 which was used to suggest the presence of building blocks that compose spatiotemporal chaos @xcite .
computations of the fractal dimension have also been conducted for experimentally accessible systems such as rayleigh - bnard convection ( the buoyant convection that occurs in a shallow fluid layer heated from below ) .
extensive chaos was found in large periodic boxes @xcite and in finite cylindrical domains @xcite yielding a natural chaotic length scale @xmath15 corresponding approximately to the width of a convection roll pair
. however , such calculations are very expensive and it remains unclear if a fluid system such as rayleigh - bnard convection is microextensive . the approach we take is to study the dynamics of a periodic lattice of coupled variables ( discrete in space and continuous in time ) whose equations of motion were originally developed by edward lorenz in 1996 with the dynamics of the atmosphere and fluid convection in mind @xcite .
the model has since become important in the study of forecasting spatiotemporally chaotic systems .
the model is now known as the lorenz-96 model and is described in detail below .
we compute the spectrum of lyapunov exponents and fractal dimension using standard approaches .
the contribution of this paper is a systematic and careful study of the variations in the fractal dimension with changes in system parameters to gain physical insights into high - dimensional chaos . by exploring a simple model we are able to perform very long - time simulations , over many initial conditions , and for a broad range of system parameters
the lorenz-96 model is a simple model developed by edward lorenz in 1996 to study difficult questions regarding predictability in weather forecasting @xcite .
the model is constructed of variables that represent the continuous time variation of an atmospheric quantity of interest , such as temperature or vorticity , at a discrete location on a periodic lattice representing a latitude circle on the earth .
the discrete variables are coupled spatially and their equation of motion includes contributions relevant to fluid systems including a quadratic nonlinearity , dissipation , and a constant external forcing .
however , the equations are phenomenological and can not be derived systematically from a more rigorous description . despite these simplifications
the lorenz-96 model has emerged as an important and often used model system for the testing of new ideas in the atmospheric sciences @xcite and in the general study of spatiotemporal chaos @xcite . in our work
, we explore the lorenz-96 model as a numerically accessible model with phenomenological relevance to fluid systems .
the computational cost of a systematic study of the variation of the fractal dimension with system parameters is significant .
it is anticipated that our exploration of a simple model will yield new insights that can be used to guide future studies of more complicated systems .
mathematically , the lorenz-96 model is a linear lattice of @xmath2 variables where the dynamics of the @xmath16th variable is given by @xmath17 for @xmath18 . in this equation @xmath19
represents an external driving , @xmath20 is a damping term , and the quadratic term is an advection term that has been constructed to conserve kinetic energy ( represented as the sums of squares of @xmath21 ) in the absence of damping .
we consider the case of constant external forcing @xmath19 and where the the lattice is periodic in space .
although other boundary conditions are possible , periodic boundaries are of particular relevance for the atmospheric systems of which this model was intended to describe .
we will show that the chaotic and periodic solutions we explore are composed of traveling structures that travel completely around the periodic - lattice many thousands of times . in light of this ,
other boundary conditions such as an absorbing boundary , would have a very strong impact on the dynamics .
we have not explored these possibilities in detail .
the two parameters @xmath2 and @xmath19 completely determine the dynamics .
the value of @xmath2 is the system size and @xmath19 is the magnitude of the external forcing . in the following we are interested in the variation of the fractal dimension with these parameters .
we explore system sizes @xmath22 since @xmath23 does not yield interesting dynamics due to the nature of the coupling . for small values of the forcing @xmath24
it has been shown @xcite that all solutions decay to the steady solution @xmath25 .
we compute the spectrum of lyapunov exponents @xmath26 and the fractal dimension @xmath5 using the standard procedure described in detail in ref .
there are @xmath2 exponents and for each exponent a set of equations linearized about eq .
( [ eq : model ] ) are evolved simultaneously to yield the dynamics of perturbations arbitrarily close to the full nonlinear system .
these tangent space equations are @xmath27 where @xmath28 is the @xmath29th perturbation about lattice site @xmath16 and @xmath1 .
the perturbations are reorthonormalized using a gram - schmidt procedure after a time @xmath30 to yield the magnitude of their growth @xmath31 .
each reorthonormalization yields a value of the instantaneous lyapunov exponent @xmath32 this is repeated and the average value of @xmath33 yields the finite time lyapunov exponent @xmath34 where @xmath35 is the number of reorthonormalizations performed .
the limit @xmath36 yields the infinite - time lyapunov exponent . in our numerical simulations , we begin from random initial conditions and use a fourth - order runge - kutta time integration with a time step @xmath37 .
typically , we integrate forward for @xmath38 time units before starting the calculation of lyapunov exponents to ensure that all transients have decayed . at this point
we begin integrating the tangent space equations using @xmath39 .
all of our reported results are for very long simulation times @xmath40 and we have computed results for each set of parameter values for 10 to 50 different random initial conditions .
we first explore the dynamics for a small value of the forcing term , @xmath41 , over a wide range of system sizes , @xmath42 .
we find that the dynamics are characterized by windows of periodic and chaotic behavior .
figure [ fig : space - time ] shows space - time plots for @xmath43 illustrating the variety of dynamics present . in fig .
[ fig : space - time](a ) the dynamics are periodic for @xmath44 yielding a wave of constant velocity traveling from right to left . in fig .
[ fig : space - time](b ) we show the interesting case @xmath45 where the dynamics are chaotic with a small value of the fractal dimension .
the dynamics consist of a distorted wave structure traveling from right to left . in fig .
[ fig : space - time](c ) chaotic dynamics are shown for @xmath46 .
this illustrates the typical chaotic dynamics that we have observed where the traveling wave structure is still apparent but with significant distortions and deviations .
it is evident from these space - time plots that the time for a wave structure to travel completely around the periodic lattice ring is @xmath47 time units for sizes of the lattice rings used here .
our typical simulation time is on the order of @xmath48 time units which indicates the number of complete rotations in one of our simulations is approximately @xmath49 .
the very long simulation times were required to gather statistics with sufficient accuracy to address many of the subtle questions we study . despite the simple nature of the model studied the requirement for very long - time simulations is significant .
overall , this suggests that the slow and noisy nature of the convergence of the lyapunov spectrum can pose significant computational challenges for more complicated models .
figure [ fig : f5](a ) illustrates the variation in dynamics with system size .
the circles represent system sizes yielding periodic dynamics and the ordinate is the magnitude of the period duration . for systems @xmath50
the period of oscillation is approximately 2 time units .
the triangles represent system sizes that yield chaotic dynamics . in order to combine all of the results on a single plot the chaotic states
were arbitrarily assigned a period of ` @xmath51 ' .
for each system size @xmath2 we performed 10 long - time numerical simulations starting from different random initial conditions .
the type of dynamics found was independent of the initial conditions used .
the size of the windows of chaotic dynamics are largest for the smaller system sizes .
for @xmath52 the chaotic solutions appear in windows of 4-lattice spacings with the occurrence of a single 5-lattice window yielding an average size of @xmath53 .
the chaotic dynamics are separated by windows of periodicity of 1-lattice spacing with one occurrence of a 2-lattice window to give an average size of @xmath54 .
[ c]l@l@l@l@l & @xmath14 & @xmath55 & @xmath56 + @xmath57 & 2.6 & 5.7 & 5.2 + @xmath58 & 1.35 & 2.0 & 4.5 figure [ fig : f5](b ) shows the variation of @xmath5 with system size @xmath2 .
the circles represent the fractal dimension for system sizes yielding chaos and the solid line is to guide the eye .
the error in @xmath5 at each value of @xmath2 is @xmath59 as determined by the standard deviation in the fractal dimension from simulations initiated with 10 different initial conditions .
the dashed line is a linear curve fit through the data for @xmath52 to yield an estimate of @xmath60 describing the dimension of extensive chaos . in all of our calculations of @xmath5 we use a 3rd order polynomial curve fit to determine an accurate value for the number of lyapunov exponents that must be included for the sum to vanish ( the curve fit uses only the 4 sums with the values closest to zero ) .
the arrows highlight the gaps between chaotic dynamics indicating system sizes yielding periodic dynamics . using the slope of the dashed line of fig .
[ fig : f5](b ) and eq .
( [ eq : chaotic_length ] ) the chaotic length scale is @xmath61 . for reference
, the characteristic lengths determined from our calculations are collected in table [ table : char - lengths ] .
therefore the size of the windows containing chaos are @xmath62 and the size of the windows containing periodic dynamics are @xmath63 . to quantify the deviations of the dimension from purely extensive chaos we define , @xmath64 the overall maximum value of the deviation from extensivity occurs for @xmath45 where @xmath65 .
if one only considers @xmath66 the maximum deviation is @xmath67 . to estimate the error in these calculations we computed results at each value of @xmath2 using 10 different random initial conditions .
the standard deviation of the values of @xmath5 is @xmath68 which is too small to include as error bars in fig .
[ fig : f5](b ) . the deviations from extensivity exhibit regular variations about the line of purely extensive chaos for @xmath69 . to quantify the length scale of these deviations
we fit a curve through the data symbols to determine the average wavelength of these deviations which we will denote as @xmath55 . using this approach yields
if each wavelength @xmath55 contains a pair of degrees of freedom this yields @xmath71 lattice spacings for the average volume of a single degree of freedom .
it is interesting to point out that @xmath72 suggesting that the deviations from extensivity are due to the addition of new chaotic degrees of freedom as the system size is increased .
this is similar to what has also been observed for the complex ginzburg - landau equation @xcite . [
cols="^,^ " , ] for @xmath41 .
( a ) periodic dynamics ( circles ) where the ordinate indicates period duration and chaotic dynamics ( triangles ) .
chaotic dynamics have been assigned an arbitrary period of ` @xmath51 ' to fit on a single plot .
( b ) deviations from extensivity illustrated by the variation of @xmath5 with @xmath2 .
the circles are simulation results and the solid lines are to guide the eye .
the maximum value of @xmath65 occurs for @xmath45 , for @xmath52 the maximum deviation is @xmath67 .
the arrows indicate windows of periodic dynamics .
the error in these calculations is @xmath59 as determined by the standard deviation of @xmath5 using @xmath73 different initial conditions at each @xmath2 .
, title="fig:",scaledwidth=45.0% ] for @xmath41 .
( a ) periodic dynamics ( circles ) where the ordinate indicates period duration and chaotic dynamics ( triangles ) .
chaotic dynamics have been assigned an arbitrary period of ` @xmath51 ' to fit on a single plot .
( b ) deviations from extensivity illustrated by the variation of @xmath5 with @xmath2 .
the circles are simulation results and the solid lines are to guide the eye .
the maximum value of @xmath65 occurs for @xmath45 , for @xmath52 the maximum deviation is @xmath67 .
the arrows indicate windows of periodic dynamics .
the error in these calculations is @xmath59 as determined by the standard deviation of @xmath5 using @xmath73 different initial conditions at each @xmath2 . , title="fig:",scaledwidth=45.0% ] we now compare these results based upon the use of lyapunov exponents with what is found by analyzing the spatial and temporal characteristics of the patterns .
analysis of the patterns are of particular interest because these measurements are often available in experiment whereas the lyapunov exponent based diagnostics are not . figure [ fig : patterns_f5](a ) shows the variation of the correlation time @xmath74 with increasing system size .
the correlation time is computed from @xmath75 where @xmath76 is the autocorrelation of the @xmath16th variable .
the results shown are the time average value of the correlation time for the variable @xmath77 in the periodic lattice . each data point
is also averaged over simulations begun from 10 different random initial conditions .
the open symbols represent chaotic dynamics and the filled symbols represent periodic dynamics .
the results exhibit a scatter about the mean value of @xmath78 with a coefficient of variation of 9.6% ( defined as the standard deviation divided by the mean ) .
the results indicate that the time dynamics remain relatively constant as the system size increases . to quantify the spatial variation of the pattern we have computed the average pattern wavelength @xmath56 with increasing system size as illustrated in figure [ fig : patterns_f5](b ) .
the wavelength is computed in fourier space using the location of the largest peak at small wavenumber to capture the size of the basic wave structure and is averaged over all time for each initial condition .
computation of the two - point spatial correlation length is difficult due the lack of an exponential decay in the correlation functions for the parameters we have explored . for smaller systems
@xmath79 the deviations are largest whereas for @xmath80 the wavelength varies about an average value of @xmath81 lattice spacings . for the chaotic solutions , the fluctuations about the average value over the 10 different initial conditions yields a coefficient of variation of approximately @xmath82 . for the periodic dynamics the wavelength of the periodic states were nearly identical to the precision of our calculations over the different initial conditions . .
( a ) the variation of the average correlation time @xmath74 with system size @xmath2 .
the dashed line is the average value of @xmath74 .
( b ) the variation of the pattern wavelength @xmath56 with system size . in both panels
open symbols represent chaotic dynamics and filled symbols represent periodic dynamics.,title="fig:",scaledwidth=45.0% ] .
( a ) the variation of the average correlation time @xmath74 with system size @xmath2 .
the dashed line is the average value of @xmath74 .
( b ) the variation of the pattern wavelength @xmath56 with system size . in both panels
open symbols represent chaotic dynamics and filled symbols represent periodic dynamics.,title="fig:",scaledwidth=45.0% ] our space - time diagnostics indicate that with increasing system size the dynamics tend toward a state where @xmath83 and @xmath84 .
the wavelength of the pattern in the large system limit is @xmath85 indicating that each wavelength of the pattern contains approximately 2 chaotic degrees of freedom on average . in this case
, it is useful to compare the deviations from extensivity with the variations in the wavelength of the patterns for increasing system size .
the variation of @xmath56 about its mean value is quite similar to what is found for the deviations from extensivity @xmath86 as illustrated in fig .
[ fig : wavelength_dimension ] .
we plot the variation of the normalized wavelength @xmath87 and the normalized deviations from extensivity @xmath88 with the system size .
the constant normalization factors are chosen such that the normalized wavelength and deviation from extensivity equal unity for @xmath89 allowing both curves to be shown on a single plot .
the circles are results for the wavelength of the pattern where solid symbols indicate periodic dynamics , open symbols represent chaotic dynamics , and the solid line is a curve fit .
the dashed line and the square symbols represent the deviations from extensivity .
gaps in the dashed line occur for system sizes exhibiting periodic dynamics .
figure [ fig : wavelength_dimension ] illustrates a correlation between the pattern wavelength and the deviation from extensivity .
the general trend is that both the fractal dimension and the pattern wavelength increase with increasing system size until the pattern adjusts by adding an additional wave structure to the periodic lattice effectively reducing the average wavelength of the pattern . with the addition of the wave structure
the dynamics are periodic and the fractal dimension vanishes .
a similar trend is seen in the work of fishman _
et al . _
@xcite where periodic dynamics were found for system sizes where the deviation from extenisivity was predicted to be a minimum for the one - dimensional complex ginzburg - landau equation . and the normalized deviation from extensivity @xmath88 for @xmath41 and @xmath90 .
the constant normalization factor is chosen such that @xmath88 and @xmath87 each equal unity for @xmath89 allowing both curves to fit on a single plot .
the wavelength results are given by the circles and the solid line ( open symbols are chaotic dynamics , filled symbols are periodic dynamics ) .
the deviations from extensivity are shown by the square symbols and the dashed line where gaps indicate regions of periodic solutions . in both cases the lines are curve fits to guide the eye.,scaledwidth=50.0% ] we
next explore the dynamics for a larger value of the forcing @xmath91 over the range of system sizes @xmath92 .
for each system size we have performed 50 independent numerical simulations starting from different random initial conditions .
each simulation was allowed to continue until @xmath93 and in computing our results we only used data in in the time interval @xmath94 to ensure the decay of all transients and in order to gather good statistics .
we found chaotic dynamics at every system size and for each random initial condition . a representative space - time plot for our results is shown in fig .
[ fig : space - time](d ) for the case of @xmath44 .
the patterns consist of distorted wave structures traveling from right to left .
the spectra of lyapunov exponents are shown in fig .
[ fig : spec](a ) for six different values of system size .
the results are normalized by @xmath2 and for extensive chaos the spectra collapse onto a single curve as expected .
the solid line is the average of the results for the different system sizes shown .
the variation of the summation of the exponents with the number of exponents is illustrated in fig .
[ fig : spec](b ) for the case of @xmath95 .
the solid line is a 6th order polynomial curve fit through the average values of the data .
the fractal dimension is indicated by the vertical dashed line at the location where the summation vanishes . with @xmath96 for @xmath18 where @xmath2 is the system size and @xmath91 .
the solid line is the average value using all of the data .
( b ) the variation of the summation of the exponents with @xmath16 for @xmath91 and @xmath95 .
the fractal dimension @xmath5 is the value where the summation vanishes .
the solid line is a 6th order polynomial curve fit and the fractal dimension @xmath5 is shown by the vertical dashed line.,title="fig:",scaledwidth=45.0% ] with @xmath96 for @xmath18 where @xmath2 is the system size and @xmath91 .
the solid line is the average value using all of the data .
( b ) the variation of the summation of the exponents with @xmath16 for @xmath91 and @xmath95 .
the fractal dimension @xmath5 is the value where the summation vanishes .
the solid line is a 6th order polynomial curve fit and the fractal dimension @xmath5 is shown by the vertical dashed line.,title="fig:",scaledwidth=45.0% ] the variation of the fractal dimension with system size is shown in fig .
[ fig : deviations ] .
figure [ fig : deviations](a ) illustrates the variation of @xmath5 with @xmath2 where the symbols represent the average value over the 50 different initial conditions .
an estimate of the error in these calculations is the standard deviation of the values of the dimension at each system size . using this ,
the error is found to be quite small with a magnitude of @xmath97 over the entire range of system sizes .
the solid line is a linear curve fit through the data points indicating extensivitiy .
this yields a value of @xmath98 for any system size @xmath2 and using eq .
( [ eq : chaotic_length ] ) yields a value of the natural chaotic length scale of @xmath99 .
it is important to point out that since @xmath100 , incremental changes in @xmath2 by adding a single lattice site allow the variation in the fractal dimension to be observed for changes in system size that are smaller than the chaotic length scale . .
( a ) the variation of @xmath5 with @xmath2 , the solid line is a linear curve fit through the data indicating extensive chaos .
for @xmath101 the dynamics are periodic and are not included in the curve fit .
( b ) the variation of the deviation from extensivity @xmath86 with @xmath2 .
the circles are results from numerical simulation and the solid line is a curve fit to guide the eye .
the deviations from extensivity are on the order of 5% and decreases with increasing @xmath2 .
the fluctuations about the average values shown in each panel , as determined from the standard deviation of results from 50 different initial conditions , is on the order of @xmath97.,title="fig:",scaledwidth=45.0% ] .
( a ) the variation of @xmath5 with @xmath2 , the solid line is a linear curve fit through the data indicating extensive chaos .
for @xmath101 the dynamics are periodic and are not included in the curve fit .
( b ) the variation of the deviation from extensivity @xmath86 with @xmath2 .
the circles are results from numerical simulation and the solid line is a curve fit to guide the eye .
the deviations from extensivity are on the order of 5% and decreases with increasing @xmath2 .
the fluctuations about the average values shown in each panel , as determined from the standard deviation of results from 50 different initial conditions , is on the order of @xmath97.,title="fig:",scaledwidth=45.0% ] the deviations of the fractal dimension about extensivity @xmath86 are shown in fig .
[ fig : deviations](b ) where we have used eq .
( [ eq : deviation ] ) and the solid line is to guide the eye .
the magnitude of the deviation from extensivity for @xmath102 is @xmath103 at its largest value and tends towards zero with increasing @xmath2 .
error bars are included and , over the entire range of system sizes , have a magnitude on the order of @xmath97 which is 3 orders of magnitude smaller than the absolute values of @xmath86 . due to the very small magnitude of our error bars
we can discern these deviations in the fractal dimension from extensivity for all values of the system size .
this suggests that the deviations from extensivity are not entirely finite - size effects and are inherent to the underlying chaotic dynamics .
for the larger system sizes @xmath104 the magnitude of the deviations from extensivity reduce to @xmath105 .
we note that over this range the magnitude of the error is on the order of @xmath106 and these small deviations are still an order of magnitude larger than the error bars . figure [ fig : deviations](b ) illustrates that the variation of @xmath86 with @xmath2 is non - monotonic in its oscillating decay in amplitude toward extensivity .
the previous studies exhibiting deviations from extensivity of ohern _ et al .
_ @xcite and fishman _ et al . _
@xcite both observed a rapid monotonic decay towards extensive chaos . at present
we do not have a good understanding of the physical origin of the non - monotonicity in our results which is perhaps related to the discrete spatial structure of the lorenz-96 model .
it is useful to separate these findings into two regimes , a small system regime where @xmath102 and a large system regime where @xmath107 .
the small system regime includes the onset of extensivity whereas the large systems are extensive . in a more complicated system , one would typically only have access to the deviations from extensivity for the small systems sizes since the deviations are largest in this range .
however , this is also the regime where one could expect finite size effects to also be important and complicate the results .
the large system limit is expected to be less influenced by these finite size effects and to provide a better estimate of the length scale of a chaotic degree of freedom . over these two regimes
we find that the wavelength of the deviations varies significantly . to quantify this we compute the zero crossings of @xmath86 over each range to estimate the wavelength of oscillation about @xmath98 given by @xmath55 . for @xmath102
there are 3 wavelengths with values of @xmath108 that yield an average of @xmath109 . assuming each wavelength of @xmath86 corresponds to the addition of two degrees of freedom yields @xmath110 for the size of a single degree of freedom .
comparison of this length with the natural chaotic length scale yields that the size of a single degree of freedom is approximately @xmath111 .
the large spread in the values of the wavelengths suggests some competition with pattern selection mechanisms .
in the large system limit @xmath112 there are 4 wavelengths characterizing the deviations from extensivity with values @xmath113 yielding an average value of @xmath114 .
the magnitude and variation of the wavelengths are smaller than what was found for the smaller systems . using the same arguments as before yields that the size of a single degree of freedom is @xmath115 .
comparison with the natural chaotic length scale yields that a single degree of freedom is @xmath116 . as in the case of @xmath41
the results yield that @xmath117 suggesting that the natural chaotic length scale and the wavelength of the deviations from extensivity are related . .
( a ) the variation of the average correlation time @xmath74 with system size @xmath2 .
the dashed line is the average value of @xmath74 .
( b ) the variation of the average pattern wavelength @xmath56 with system size .
the solid line is a curve fit of the form @xmath118.,title="fig:",scaledwidth=45.0% ] .
( a ) the variation of the average correlation time @xmath74 with system size @xmath2 .
the dashed line is the average value of @xmath74 .
( b ) the variation of the average pattern wavelength @xmath56 with system size .
the solid line is a curve fit of the form @xmath118.,title="fig:",scaledwidth=45.0% ] in order to better characterize the patterns we have also computed characteristic time and length scales describing their dynamics .
our motivation is to provide further insight into the transition between the small and large system limits in order to separate finite size effects from dynamical effects related to extensivity .
we are also interested in quantifying any relationship between the lyapunov based diagnostics and the pattern dynamics .
figure [ fig : patterns_f10](a ) shows the variation of the average correlation time @xmath74 with @xmath2 over 50 different initial conditions .
the time dynamics are noisy and the correlation time has an average value of approximately @xmath119 ( indicated by the dashed line ) .
the time dynamics are faster than what was found for @xmath41 as expected for the increased value of the forcing term .
figure [ fig : patterns_f10](b ) illustrates the variation in the average wavelength of the patterns with @xmath2 over the different initial condtions .
the wavelength variation shows two regions of interest . for small system sizes @xmath120
the pattern wavelength increases rapidly . for larger systems
the wavelength remains relatively constant with an average value of @xmath121 .
the region of increasing wavelength corresponds to system sizes where the pattern selection is affected significantly by the size of the domain .
this corresponds roughly with the region where the fractal dimension is approaching extensivity in fig .
[ fig : deviations](b ) .
these trends are well captured by a power - law , the solid line in the figure is a curve fit through the data of the form @xmath118 .
the fluctuations of the wavelength about the mean value for these chaotic states yields a coefficient of variation of approximately @xmath122 .
these results suggest that our measurements of the deviations in the fractal dimension for the large system limit are quantifying the chaotic dynamics in a regime that is not strongly affected by finite size effects .
our space - time diagnostics indicate that the pattern dynamics , in the large system limit , have temporal variations with @xmath123 and a spatial wavelength on the order of @xmath124 . in this case , @xmath125 indicating that a wavelength of the pattern contains , on average , 3.3 chaotic degrees of freedom .
a more detailed exploration of the relationship between @xmath56 and @xmath14 for increasing values of the forcing @xmath19 will be explored in the following section . in the previous sections we have considered the large system limit , defined as the limit where the chaotic degrees of freedom are much smaller than the system size .
this was achieved by holding the external forcing @xmath19 fixed while the system size @xmath2 was increased .
this is the typical manner to study spatiotemporal chaos and has been referred to as the ` spatiotemporal chaos ' limit @xcite .
for example , in a rayleigh - bnard convection experiment this could be accomplished by holding the rayleigh number constant while increasing the aspect ratio of the convective domain
. however , it is also possible to explore the large system limit by keeping the system size fixed while increasing the external forcing .
this has been referred to as the strong driving or ` strong turbulence ' limit @xcite . in this case
, it is expected that the chaotic degrees of freedom will become smaller with increasing forcing to yield the large system limit . in a rayleigh - bnard convection experiment
this would be accomplished by increasing the rayleigh number in a convection domain of fixed size .
it has been conjectured that the fractal dimension will also exhibit a power - law dependence with respect to the value of the forcing @xcite .
we have explored this strong driving limit in the lorenz-96 model by performing a series of simulations for increasing values of @xmath19 while the system size @xmath2 is held constant .
we studied 6 values of the forcing over the range @xmath126 . for any particular value of @xmath19 we performed 10 numerical simulations starting from different random initial conditions and allowed the simulation to run for @xmath127 time units to ensure good statistics .
we computed these results for 5 different system sizes where @xmath128 .
it is more convenient to discuss these results using the intensive dimension density , @xmath129 the variation of the dimension density with external forcing is shown in fig .
[ fig : df ] .
each symbol is the average value of the 10 simulations from different random initial conditions .
the standard deviation of the results are @xmath97 and have not been included as error bars due to their small magnitude . with the forcing @xmath19 and system size @xmath2 .
the results are for 5 different system sizes indicated in the legend .
the symbols are simulation results and the solid line is a power - law curve fit through the average value of the data . each symbol is the average value from results computed using 10 different initial conditions .
the standard deviation of these results for each symbol is @xmath97.,scaledwidth=50.0% ] the dimension density exhibits power - law behavior and the results for different values of @xmath2 collapse onto a single power - law curve given by , @xmath130 where @xmath131 , @xmath132 , and @xmath133 .
the power - law curve fit is given by the solid line .
the coefficient @xmath134 in this case is the predicted value of the dimension density for external forcing of infinite magnitude @xmath135 .
the scatter in the results is quite small for all values of @xmath19 explored with the largest deviations occurring at the smallest value of the forcing .
the coefficients of the power - law are also collected in table [ table : powerlaws ] for reference .
the inverse of the dimension density is simply the natural chaotic length scale @xmath136 .
therefore fig .
[ fig : df ] can be used to illustrate the manner in which the chaotic length scale decreases with increasing forcing .
this is a spatially discrete system with @xmath2 degrees of freedom and the theoretical limit for the dimension density is @xmath137 corresponding to a chaotic length scale of a single lattice spacing @xmath138 . using our results for @xmath135 yields a value of @xmath139 for the chaotic length in the limit of infinite forcing
this is in contrast to a fluid system described by partial differential equations with an infinite number of degrees of freedom .
it is interesting to note that the variation of the fractal dimension with the degree of external forcing has been explored numerically for turbulent rayleigh - bnard convection to yield a linear dependence @xcite .
we now compare several measures of characteristic time and length scales for increasing values of the magnitude of external forcing .
both the temporal and spatial scales decrease with increasing forcing as expected .
the variation of these scales are well described by a power - law of the functional form given by eq .
( [ eq : powerlaw ] ) .
the variation of the inverse leading order lyapunov exponent and the correlation time are shown in fig [ fig : timef ] .
the symbols are results from our simulations for each value of @xmath19 .
the value of @xmath140 yields a time scale related to the predictability of the system . the solid line in fig [ fig : timef](a ) is a power - law curve fit using the values shown in table [ table : powerlaws ] .
the inverse lyapunov exponents have been scaled by the system size @xmath2 so that all data can be represented on a single plot .
the value of @xmath141 represents @xmath142 for infinite @xmath19 .
figure [ fig : timef](b ) illustrates the variation of the average correlation time @xmath74 with increased forcing and over the 10 different initial conditions for a large system with @xmath143 .
the solid line is the power - law curve fit using the values of table [ table : powerlaws ] .
the fluctuations of the correlation time about the mean value over the different initial conditions is quite noisy and yields a coefficient of variation of 7.7% . in the limit of infinite forcing the average correlation time @xmath144 .
our results suggest that the predictability decreases faster than the correlation time indicating the possibility of at least two different time scales . .
( a ) the variation of the normalized inverse of the leading order lyapunov exponent @xmath142 .
data are shown for 5 different system sizes as indicated in the legend .
the solid line is a curve fit through the data of the form @xmath145 .
( b ) the variation of the average correlation time @xmath74 with @xmath19 for a system size of @xmath143 .
the symbols are results from simulation and the solid line is the power - law @xmath146 .,title="fig:",scaledwidth=45.0% ] . ( a )
the variation of the normalized inverse of the leading order lyapunov exponent @xmath142 .
data are shown for 5 different system sizes as indicated in the legend .
the solid line is a curve fit through the data of the form @xmath145 .
( b ) the variation of the average correlation time @xmath74 with @xmath19 for a system size of @xmath143 .
the symbols are results from simulation and the solid line is the power - law @xmath146 .,title="fig:",scaledwidth=45.0% ] [ c]l@l@l@l & @xmath134 & @xmath147 & @xmath148 + @xmath149 & 0.93 & - 4.03 & 1.32 + @xmath142 & 0.007 & 5.9 & 2.6 + @xmath74 & @xmath51 & 0.35 & 0.56 + @xmath14 & 1.12 & 39.74 & 2.2 + @xmath56 & 4.38 & -290.8 & 3.5 + the pattern wavelength and the chaotic length scale both decrease with a power - law variation given by eq .
( [ eq : powerlaw ] ) for increasing values of the magnitude of the external forcing .
the coefficients of the power - law variation are given in table [ table : powerlaws ] .
the variation of these length scales is shown in fig .
[ fig : spacef](a ) .
the pattern wavelength is represented using square symbols and the natural chaotic length scale is represented using circle symbols .
the length scales have been normalized by their magnitude at @xmath41 to facilitate a comparison using a single plot .
the normalized length scales are referred to as @xmath87 and @xmath150 , respectively .
the coefficient of variation , over the different initial conditions , for the chaotic length scale is quite small @xmath151 whereas the pattern wavelength measurements are quite noisy with a coefficient of variation of @xmath152 .
it is clear that the natural chaotic length scale decreases more rapidly than the pattern wavelength .
the ratio of these two length scales yields an estimate for the number of chaotic degrees of freedom per pattern wavelength and is shown in fig .
[ fig : spacef ] . for small values of the forcing
there are approximately 2 degrees of freedom per wavelength which increases to nearly 4 degrees of freedom per wavelength for large values of the forcing .
the separation of these two length scales suggests that for larger values of the external forcing significant contributions to the overall disorder are from sub - wavelength structures in the pattern dynamics .
a clear signature of the sub - wavelength chaotic degree of freedom was not found using our spatial diagnostics . for a system size of @xmath143 .
( a ) the normalized natural chaotic length scale @xmath153 ( circles ) and the normalized pattern wavelength @xmath87 ( squares ) .
the normalization factors used are the respective values of @xmath14 and @xmath56 at @xmath41 .
the power - law curve fits for the unscaled length scales are @xmath154 and @xmath155 .
( b ) the ratio @xmath156 yields an estimate for the number of chaotic degrees of freedom per pattern wavelength.,title="fig:",scaledwidth=45.0% ] for a system size of @xmath143 .
( a ) the normalized natural chaotic length scale @xmath153 ( circles ) and the normalized pattern wavelength @xmath87 ( squares ) .
the normalization factors used are the respective values of @xmath14 and @xmath56 at @xmath41 .
the power - law curve fits for the unscaled length scales are @xmath154 and @xmath155 .
( b ) the ratio @xmath156 yields an estimate for the number of chaotic degrees of freedom per pattern wavelength.,title="fig:",scaledwidth=45.0% ]
using a phenomenological model relevant to fluid convection and the atmosphere we have shown that the variation of the fractal dimension with system parameters can provide physical insights into fundamental features of high - dimensional chaos .
we have used the fractal dimension to provide an approximate value of the number of chaotic degrees of freedom in the system and to estimate length scales describing the underlying chaotic dynamics . if one considers only the space - time diagnostics of the correlation time and pattern wavelength the results exhibit significant fluctuations despite the use of very long - time simulations for numerous initial conditions .
this is in contrast to what is found when using the fractal dimension and suggests the possibility of features of the dynamics that have yet to be quantified in detail .
our results indicate that the length scale describing the deviations from extensivity is approximately equal to the natural chaotic length scale which suggests the possibility of important spatial structures that contribute significantly to the chaotic dynamics .
clearly identifying such features , and relating them to experimentally accessible quantities , remains an important open challenge in the characterization of systems driven far - from - equilibrium . for systems of small size with low values of external forcing
we find very complicated dynamics with windows of periodic and chaotic dynamics .
this is similar to what has been found both numerically @xcite and experimentally @xcite in fluid systems such as rayleigh - bnard convection . for intermediate forcing ,
extensive chaos emerges with the important feature of significant deviations from extensivity for incremental changes in system size .
our results suggest that such behavior may be present in experimentally accessible fluid systems .
the ratio of our measured length scales , such as the ratio of the chaotic length scale @xmath14 to the wavelength of the deviations from extensivity @xmath55 do not yield integer values as found for the 1d complex ginzburg - landau equation @xcite .
this is perhaps also due in part to the discrete spatial nature of the lorenz-96 model .
the variation of the fractal dimension with external forcing yields insights into the manner in which the chaotic length scale decreases as the strong driving limit is approached . in the case of increasing system size while holding the forcing fixed
, extensive chaos occurs when the exponent relating the dimension and system size is equal to the number of spatially extended directions .
however , a similar theoretical understanding of the growth in the fractal dimension with increased forcing remains an open challenge .
the fact that our results are independent of system size is promising and suggests that perhaps this is an underlying feature of some generality .
it is anticipated that our results will be useful in guiding future efforts to explore the extensive chaos of experimentally accessible systems . | we explore the high - dimensional chaotic dynamics of the lorenz-96 model by computing the variation of the fractal dimension with system parameters .
the lorenz-96 model is a continuous in time and discrete in space model first proposed by edward lorenz to study fundamental issues regarding the forecasting of spatially extended chaotic systems such as the atmosphere .
first , we explore the spatiotemporal chaos limit by increasing the system size while holding the magnitude of the external forcing constant .
second , we explore the strong driving limit by increasing the external forcing while holding the system size fixed . as the system size is increased for small values of the forcing we find dynamical states that alternate between periodic and chaotic dynamics .
the windows of chaos are extensive , on average , with relative deviations from extensivity on the order of 20% . for intermediate values of the forcing we find chaotic dynamics for all system sizes past a critical value .
the fractal dimension exhibits a maximum deviation from extensivity on the order of 5% for small changes in system size and decreases non - monotonically with increasing system size .
the length scale describing the deviations from extensivity and the natural chaotic length scale are approximately equal in support of the suggestion that deviations from extensivity are due to the addition of chaotic degrees of freedom as the system size is increased . as the forcing is increased at constant system size
the fractal dimension exhibits a power - law dependence .
the power - law behavior is independent of the system size and quantifies the decreasing size of chaotic degrees of freedom with increased forcing which we compare with spatial features of the patterns . |
complex ( dusty ) plasma is a low - temperature plasma including microparticles . due to electron absorption , particles acquire a considerable electric charge .
thus , a strongly coupled coulomb system is formed @xcite .
such plasma represents a natural system which makes it possible to observe various collective phenomena at the level of individual particles . in experimental setups ,
complex plasmas are usually studied in gas discharges at low pressures , e.g. , in radio frequency ( rf ) discharges .
a large homogeneous bulk of complex plasma , which almost fills the entire discharge volume , can be observed under microgravity conditions either in parabolic flights @xcite or onboard the international space station ( iss ) @xcite . in recent studies ,
attention was focused on tracer particles or projectiles moving through a cloud of complex plasma .
such projectiles are generated using controlled mechanisms of acceleration @xcite , or they can appear sporadically @xcite .
projectiles moving with supersonic velocity lead to the formation of extended mach cones ; subsonic ( slow ) ones produce localized disturbances of surrounding particles . in ref .
, it was suggested that the latter regime , realized when a relatively large subsonic projectile moves in a dense cloud of smaller particles , can be well approximated by a flow of an incompressible fluid . in the framework of the same hydrodynamic approximation , it was demonstrated that , along with the regular neutral gas drag , there is an additional force exerted on the projectile due to friction between neutral atoms and the particle fluid @xcite . in this study
, we try to estimate the deformation threshold for an ( initially spherical ) cavity around a subsonic projectile .
this deformation emerges abruptly as the projectile velocity increases , as it is seen in snapshots given in @xcite . to solve this problem
, we employ the navier stokes equation for a compressible fluid taking into account friction between dust particles and atoms of neutral gas .
we do not treat the deformation self - consistently .
instead , we imply that the projectile velocity is below the threshold value , so that a regular flow around a spherical cavity with no stall can be treated .
the solutions shows that due to friction , the pressure of a dust cloud at the boundary of the cavity behind the projectile can become negative , which entails the formation of a microscopic void free from dust particles , i.e. , the cavity deformation .
this occurs at some threshold velocity which decreases with increasing cavity size .
measurement of such velocity would make it possible to estimate the static pressure inside the dust cloud .
the paper is organized as follows . in sec .
[ s2 ] , we solve the non - stationary navier
stokes equation for an incompressible particle fluid . in sec .
[ s3 ] , the gas dynamics problem is solved and the corrections for fluid compressibility to the velocity and pressure fields are calculated .
calculation results are compared with available experimental data in sec .
[ s4 ] and the results of this study are summarized in sec .
consider an irrotational flow of incompressible particle fluid formed by the dust crystal melted around a projectile moving with the velocity @xmath0 relative to the dust , where @xmath1 is the time .
the parameter of interaction between dust particles is large due to their low kinetic energy at room temperature .
the corresponding parameter for the interaction between the projectile and dust particles can be defined as the ratio of characteristic coulomb energy to the kinetic energy of a dust particle , @xmath2 ( where @xmath3 and @xmath4 are the charges of the projectile and dust particle , respectively , in units of the electron charge @xmath5 , @xmath6 is the plasma ( ion ) screening length , and @xmath7 is the dust particle mass ) . under typical conditions , this parameter is great , @xmath8 @xcite .
in such a strongly coupled system , each particle ( including the projectile ) finds itself in the center of the spherically symmetric wigner seitz cell .
since each cell is electrically neutral , the interaction of particles can be effectively reduced to that of cell surfaces , which can be modeled by hard spheres .
this applies to the surface of a cell around the projectile as well .
this approach is valid until the cavity deformation occurs ; however , we will confine ourselves to the treatment of spherically symmetric cavities below the deformation threshold .
we denote the radius of a cavity around the projectile by @xmath9 .
the navier stokes equation describing the velocity field @xmath10 in the reference frame of the projectile has the form @xmath11 where @xmath12 is the mass density of dust fluid ( assumed in this section to be constant ) , with @xmath7 and @xmath13 being , respectively , the mass and number density of dust particles of the radius @xmath14 .
furthermore , @xmath15 is the dust pressure field , @xmath16 is the friction coefficient with @xmath17 being the accommodation coefficient @xcite , and @xmath18 , @xmath19 , @xmath20 , and @xmath21 are the mass , number density , temperature , and thermal velocity of neutral gas molecules , respectively . equation ( [ e1 ] ) assumes also that far from the projectile , dust particles are quiescent relative to neutral gas .
it was shown in @xcite that the approximation of nonviscous flow results in a fairly good description of trajectories of individual dust particles ; the estimate of the viscosity term for the dust particle fluid is indicative of the fact that in most cases , it is small @xcite .
this allowed us to omit it in eq .
( [ e1 ] ) and to confine ourselves to the nonviscous approximation . for an incompressible fluid ,
the continuity equation is reduced to @xmath22 the boundary conditions for ( [ e1 ] ) and ( [ e2 ] ) are @xmath23 , where @xmath24 , and @xmath25 at @xmath26 . for an irrotational flow ( @xmath27 ) , we substitute @xmath28 in eq .
( [ e2 ] ) to obtain the equation @xmath29 which has the solution @xcite @xmath30 or @xmath31 -{\bf u}. \label{e5}\ ] ] the pressure field @xmath32 can then be found by substituting velocity ( [ e5 ] ) in eq .
( [ e1 ] ) .
this yields the following pressure distribution at the spherical surface of a cavity @xmath33 , @xmath34 + { \displaystyle{{\rho r } \over
2}}\dot{\bf u}\cdot{\bf n } , \label{e8}\ ] ] where @xmath35 is the static pressure of dust at @xmath26 .
consider the deformation of a cavity around a projectile propagating in the dust crystal ( fig .
[ f1 ] ) .
= 0.24pt ( 500,700 ) ( -90,-40)deformation of a cavity behind moving projectile ( large bullet in the center ) moving with the velocity @xmath36 through the dust crystal ( small bullets around ) in a carrier gas . the projectile is surrounded by deformed cavity with initial radius @xmath9.,title="fig:",width=240 ] here , we will neglect finite fluid compressibility corresponding corrections are discussed in sec .
deformation can be caused solely by the formation of a void in the stream of dust fluid because the interaction parameter @xmath37 is large .
since no evidence for an appreciable attraction between dust particles has been reported in the literature , we model the cloud of dust particles by a system of hard spheres .
in such a system , a cavity can be deformed by void formation if the condition @xmath38 is satisfied over some area around a cavity .
this condition corresponds to cavitation in a metastable fluid and to stall in gas dynamics .
let us first consider the case @xmath39 . according to eq .
( [ e8 ] ) , friction between dust particles and neutral gas gives rise to a pressure increase in front of the projectile and a decrease behind it , so that @xmath32 reaches a maximum at the `` front pole '' ( where @xmath40 ) . by introducing the cosine of the polar angle , @xmath41
, we readily derive @xmath42 where @xmath43 . if @xmath44 then the pressure reaches a minimum at the `` rear pole '' ( @xmath45 ) , where @xmath46 .
this corresponds to the low - velocity regime @xmath47 , when the minimum pressure decreases ( approximately ) linearly both with @xmath36 and @xmath9 .
the cavity deformation threshold is defined by the condition @xmath48 and is reached easier for larger projectiles , at the critical velocity @xmath49 [ here we assume @xmath50/8 ] .
note that in the high - velocity regime @xmath51 the pressure minimum is shifted to a certain latitude in the rear hemisphere , approaching the `` equator '' ( @xmath52 ) asymptotically .
in this case friction plays a minor role and the critical velocity is @xmath53 .
note that for arbitrary acceleration ( assuming @xmath54 ) , one can introduce the effective damping rate @xmath55 which can have either sign , so that the critical cosine @xmath56 can have either sign as well .
thus , if a projectile experiences sufficiently strong deceleration ( @xmath57 ) , the cavity can also be deformed at the front hemisphere .
the critical velocity @xmath58 calculated above defines the emergence of a point or circle of zero pressure on the cavity surface depending on the value of @xmath59 .
it is well - known in gas dynamics that the existence of small local regions of negative pressure would not lead to a significant deformation of flow lines and , therefore , to a stall .
instead , stall emergence requires an extended region of negative pressure that covers a significant part of `` rear hemisphere '' .
there does not seem to be any criterion that would tell us how large the negative region should be to initiate a stall .
nevertheless , one can hope that the velocity @xmath58 derived above provides a reasonable lower bound estimate for the cavity deformation onset . strictly speaking ,
the theory discussed above is not applicable for the case @xmath60 because a streamlined cavity is already deformed . at the same time below the stall threshold
, this deformation is small , i.e. , the cavity is still almost spherical .
therefore , one could extend the analysis to the region of the velocities higher than the critical one and , respectively , to the region of negative pressure larger than a point or a circle on the cavity surface . in the next section ,
obtained results are generalized to the case of the finite compressibility of a fluid .
in this section , we will estimate the effect of finite compressibility of the dust cloud on the velocity and pressure fields . here , we will treat a steady ( @xmath61 ) and irrotational ( @xmath62 ) flow , so that the fluid density is a function solely of the radius - vector : @xmath63 . since @xmath62 and @xmath64 , where @xmath65 is the sound velocity , we can rewrite eq .
( [ e1 ] ) in the form @xmath66 in this case , the continuity equation includes the fluid density , @xmath67 here , @xmath68 is assumed to be constant independent of the local velocity and plasma state parameters .
we substitute @xmath69 from ( [ e14 ] ) in ( [ e15 ] ) to derive @xmath70 we will search for the solution of eq .
( [ e16 ] ) in the form @xmath71 , where @xmath72 is the solution of incompressible problem ( for @xmath73 const , i.e. , for @xmath74 ) given by eq .
( [ e5 ] ) , the respective pressure distribution @xmath75 is presented by eq .
( [ e8 ] ) ( for @xmath39 ) . obviously , @xmath76 .
we assume that @xmath77 and retain solely zeroth - order terms on the rhs of ( [ e16 ] ) to obtain @xmath78 here , we took into account that @xmath79 and hence @xmath80 .
since @xmath81 , @xmath82 , and @xmath83 is small for subsonic motion , the estimate @xmath84 following from ( [ e18 ] ) justifies our assumption .
it is seen from this estimate that both sides of eq .
( [ e18 ] ) are of the order @xmath85 , and terms other that zeroth - order ones would lead to higher - order terms in this parameter .
obviously , @xmath86 at @xmath87 .
we introduce the scaled coordinate @xmath88 and the dimensionless potential @xmath89 defined by the relation @xmath90 ( in so doing , the condition @xmath91 is satisfied ) to rewrite ( [ e18 ] ) in the dimensionless form @xmath92 where @xmath93 are the legendre polynomials and @xmath94 .
here and in what follows , differentiation with respect to scaled coordinates is assumed .
the boundary conditions for eq .
( [ e19 ] ) follow from the conditions @xmath95 and @xmath96 , i.e. , @xmath97 solution of eq .
( [ e19 ] ) is derived in appendix [ sa ] .
the resulting finite - compressibility corrections to the potential and velocity at the cavity surface follow from ( [ e29 ] ) : @xmath98 as it could be expected from symmetry considerations , @xmath99 at @xmath100 , i.e , at both poles . note
that result ( [ e21 ] ) validates the estimate @xmath101 given above . on the basis of velocity correction ( [ e21 ] ) and using eq .
( [ e14 ] ) , it is a straightforward matter to calculate a compressibility correction to the pressure @xmath102 .
this is performed in appendix [ sb ] .
it is worth mentioning that in the general case , correction to the pressure ( [ e33 ] ) includes solely even powers of @xmath103 , so that corresponding correction to an additional force exerted on the projectile due to dissipation in the surrounding particle fluid [ eq . ( 10 ) in ref .
@xcite ] vanishes . in the friction - dominated regime ( large @xmath104 )
, one can retain in ( [ e33 ] ) solely the term proportional to @xmath105 , which is independent of @xmath103 , so that the total pressure is @xmath106 thus , the pressure increases due to finite compressibility , so that the threshold velocity of cavity deformation increases as well . on the other hand , one can see that the resulting correction is small when @xmath107 .
however , we will see that the finite compressibility correction is always small for large projectiles even if this condition is not satisfied .
now we repeat the analysis of sec .
[ s2 ] for eq .
( [ e22 ] ) instead of ( [ e8 ] ) .
since the finite compressibility correction is independent of @xmath103 , the expression for @xmath108 remains the same as in sec .
consider first the case @xmath109 , when we have to set @xmath110 in eq .
( [ e22 ] ) to find the threshold velocity corresponding to emergence of a point of zero pressure at the rear pole " of a cavity : @xmath111 where @xmath112 . for the case of large cavities ( @xmath113 ) typical for treated case , approximation ( [ e001 ] ) is valid provided that the condition @xmath114 , which applies to experiments analyzed in sec .
[ s4 ] , is satisfied even if @xmath115 .
thus , finite compressibility seems to have no significant effect on the critical velocity .
the reciprocal dependence @xmath116 has the form @xmath117 upon further increase of the projectile velocity @xmath36 , the region of negative pressure on the surface of a cavity is expanded .
it is a straightforward matter to demonstrate that extrapolation of the theory to the extended region of negative pressures ( where it is formally invalid ) results in a very fast increase of this region with the increase of @xmath36 .
therefore , @xmath58 is likely to be a good estimate for the stall velocity @xmath118 . at the same time
, @xmath119 means that if we equate @xmath120 we obtain from ( [ e040 ] ) an overestimated @xmath121 , which can be treated as _ an upper bound estimate _ for the static pressure . in the case
@xmath122 typical for small projectiles , a circle with zero pressure first emerges at the critical velocity @xmath123 here , the approximation corresponding to the limit @xmath124 also coincides with that for an incompressible fluid . if @xmath58 is known from experiment , one can derive an upper bound estimate for the static pressure , @xmath125 in this case at @xmath60 , the circle expands to a region between two circles on the surface of a cavity . for small projectiles , one can also demonstrate that a fast expansion of the negative region takes place as @xmath36 is increased .
thus , @xmath58 also yields a reasonable order of magnitude estimate for a stall threshold .
based on developed theory we will analyze the possibility of cavity deformation in experiments . consider first the experiments carried out in the pk-3 plus laboratory onboard the iss under microgravity conditions .
details on the setup can be found in @xcite .
dust particles injected into the main plasma with dispensers formed a cloud around the center of the chamber . some larger particles present in the chamber as well get sporadically accelerated and penetrate into the cloud , thus forming projectiles @xcite .
consider the experiment with argon as a carrier gas .
for a gas pressure and temperature of 10 pa and @xmath126 , respectively , we have @xmath127 , @xmath128 , and @xmath129 .
a dust cloud was formed by melamine - formaldehyde particles with the radius @xmath130 and mass @xmath131 @xcite , therefore , @xmath132 .
successive frames of the path of a projectile recorded by a high - resolution camera are shown in fig .
[ f4 ] .
= 0.24pt ( 500,750 ) ( -490,-150)positions of dust particles and the projectile ( negative images ) at successive instants .
time interval between individual frames is 0.08 s. frames ( a ) and ( b ) illustrate a spherical cavity ; ( c ) and ( d ) , deformed cavity .
the projectile is accelerated from ( a ) to ( d).,title="fig:",width=432 ] the projectile velocity increases slowly from @xmath133 to @xmath134 .
since the sound velocity measured in this experiment amounts to 2.2 cm / s @xcite , such motion is subsonic .
the peculiarity of this motion is an abrupt change of cavity configuration . at the beginning
, the projectile moves in the center of a cavity [ cf .
[ f4](a ) and ( b ) ] .
then , at the point of trajectory characterized by the velocity @xmath135 , acceleration @xmath136 , and the cavity radius @xmath137 , the projectile shifts abruptly from the cavity center to its front side [ fig . [ f4](c ) ] . in the course of further motion , cavity deformation is preserved [ fig .
[ f4](d ) ] .
one can assume that at the point shown in fig .
[ f4](c ) , the velocity of the accelerating projectile exceeds a stall threshold . under specified experimental conditions , @xmath138 , hence we substitute @xmath139 in ( [ e004 ] ) to derive an upper bound estimate @xmath140 ( or @xmath141 ) . note that the correction for acceleration to the total pressure @xmath142 [ eq . ( [ e8 ] ) ] is small due to a small projectile acceleration at selected point ( @xmath143 ) and can be neglected .
the estimate obtained above can be compared with a theoretical estimation for the static pressure , which is based on dimensionality considerations . since dimensionalities of the pressure and energy density coincide , we can write with an order of magnitude accuracy @xmath144 , where @xmath145 is the wigner seitz radius for a dust crystal and @xmath13 is the number density of dust particles . with @xmath146 and @xmath147 @xcite ,
this estimation yields @xmath148 , which correlates with the upper bound estimate resulting from experiment .
microgravity conditions can also be maintained on parabolic flights .
such experiments were performed using the impf - k2 chamber @xcite .
the experiment was carried out with argon at 30 pa and @xmath149 ( @xmath150 and @xmath151 ) .
projectiles were accelerated by a special device ( cogwheel ) up to both supersonic and subsonic velocities . a typical image illustrating the motion of a subsonic projectile inside a dust cloud is shown in fig . [ f5 ] .
= 0.24pt ( 500,750 ) ( -475,-240)deformed cavity around a large subsonic projectile moving with the velocity @xmath152 .
reproduction of an enlarged fragment of fig .
3a in ref .
@xcite.,title="fig:",width=470 ] both the projectile shift from the cavity center toward its front side and the nonspherical ( elongated ) shape of a cavity are visible .
unfortunately , a point of transition from a spherical to a deformed cavity was not detected in this experiment .
nevertheless , we can estimate the stall threshold for this case . note that here , the estimate @xmath153 with @xmath154 and @xmath155 @xcite yields @xmath156 , which is close to the estimate for experiment @xcite discussed above .
therefore , we can assume that @xmath121 are close in both systems and use the upper bound estimate ( [ e001 ] ) to calculate @xmath58 for the treated system . with the cavity radius roughly estimated as @xmath157
, we derive @xmath158 , so that the projectile moves in the low - velocity regime ( @xmath159 ) .
therefore , the stall threshold for this system does not exceed 1 cm / s .
this result correlates with the projectile velocity in fig .
[ f5 ] , @xmath152 , which is still lower than the sound velocity @xmath160 @xcite .
thus , the main notion of the present study , namely that the cavity around a moving projectile can be deformed at _ subsonic velocity _ , is corroborated .
we have investigated the possibility of the deformation of a cavity around large projectiles slowly moving inside a cloud of small dust particles .
this cavity is formed due to a strong coulomb repulsion between the projectile and dust particles .
since the parameter @xmath161 characterizing this interaction is very large , cavity deformation occurs solely due to the formation of a void adjacent to the initially spherical cavity .
this is similar to the process of void formation in a fluid under negative pressure .
since we approximate the dust cloud by a system of hard spheres , zero pressure is sufficient for void formation leading to cavity deformation .
we model a collective particle subsonic motion in a complex plasma by nonviscous irrotational hydrodynamic motion of dust fluid and show that , in this model , the reason for cavity deformation is the friction between dust particles and neutral atoms of a carrier gas .
it is this friction that stipulates the emergence of a zone of negative pressure over the back of a cavity surface .
if such a zone covers a substantial part of the surface , stall takes place , which is observed as cavity deformation .
we have found that the deformation is more likely to be observed for larger projectiles .
the deformation occurs when the projectile velocity exceeds some threshold of subsonic velocity . since typical projectile velocities are of the same order of magnitude as the sound velocity
, we have studied the effect of finite compressibility of the dust fluid on velocity and pressure fields . toward that end
, we have obtained an exact solution of linearized gas dynamics equations for compressible nonviscous fluid flowing about the cavity . in so doing ,
we included both the convective and friction terms in the navier stokes equation .
calculations showed that corrections for finite compressibility are rather small .
generally , account of the compressibility leads to a shift of the area covered by a negative pressure zone .
analysis of available experimental data has validated our approach .
it was found that the cavity is deformed when either the velocity @xcite or the size @xcite of a projectile is sufficiently large .
it is worth mentioning that the phenomenon of cavity deformation is similar to the formation of a void behind a cluster of smaller particles moving in the cloud of larger ones @xcite . for intermediate projectile sizes , a transition from a spherical to a deformed cavity
was observed along the trajectory of an individual projectile .
the transition threshold is in accordance with the developed theory .
this made it possible to estimate the static pressure inside a dust cloud and , hence , opened up a possibility to use projectiles as a diagnostic tool for the dust equation of state . in this study , we restricted ourselves to an investigation of the threshold of cavity deformation and did not address the shape of the deformed cavity or displacement of the projectile away from the center of its cavity .
this problem would require an adequate account of the local nonspherical charge distribution around the projectile .
it is also possible that the observed shift of the projectile from the center of a cavity is noticeably favored by an additional drag force exerted on the projectile @xcite .
the solution of eq .
( [ e19 ] ) with boundary conditions ( [ e20 ] ) is represented in the form @xmath162 where @xmath163 are the coefficients of expansion in the legendre polynomials @xmath93 .
we substitute ( [ e24 ] ) in ( [ e19 ] ) taking into account that @xmath164 to reduce eq . ( [ e19 ] ) to a set of ordinary differential equations in spherical coordinates for nonzero functions @xmath163 , @xmath165 where @xmath166 , @xmath167 , @xmath168 , @xmath169 .
the boundary conditions for ( [ e25 ] ) follow from ( [ e20 ] ) : @xmath170 the rhs of ( [ e25 ] ) has the form @xmath171 , where @xmath172 are constants entering the functions @xmath173 .
solution of ( [ e25 ] ) is then given by @xmath174 where @xmath175 is defined by the second boundary condition ( [ e26 ] ) , whence it follows that @xmath176 finally , we obtain @xmath177
the pressure correction due to finite compressibility can be found using the corresponding velocity correction , eq .
( [ e21 ] ) . we rewrite eq .
( [ e14 ] ) in the form @xmath178 where @xmath179 and @xmath180 . by integrating eq .
( [ e30 ] ) we derive @xmath181 where @xmath182 and the terms up to @xmath183 are retained ( for simplicity we assume @xmath184 const ) .
we write the equation of state for the dust fluid in the form @xmath185 , which yields @xmath186 the first three terms of eq .
( [ e32 ] ) represent eq .
( [ e8 ] ) with @xmath187 .
hence , by substituting in eq .
( [ e32 ] ) the expressions for @xmath188 and @xmath189 [ eqs .
( [ e4 ] ) and ( [ e5 ] ) ] as well as for @xmath190 and @xmath191 [ eq . ( [ e21 ] ) ] taken at the cavity surface , we get @xmath192 . \label{e33}\end{aligned}\ ] ] 27ifxundefined [ 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 and , eds . ,
@noop _ _ , series in plasma physics ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) | we study the deformation of a cavity around a large projectile moving with subsonic velocity in the cloud of small dust particles . to solve this problem
, we employ the navier stokes equation for a compressible fluid with due regard for friction between dust particles and atoms of neutral gas .
the solutions shows that due to friction , the pressure of a dust cloud at the surface of a cavity around the projectile can become negative , which entails the emergence of a considerable asymmetry of the cavity , i.e. , the cavity deformation .
corresponding threshold velocity is calculated , which is found to decrease with increasing cavity size .
measurement of such velocity makes it possible to estimate the static pressure inside the dust cloud . |
accretion - powered millisecond pulsars ( amps ) provide a unique laboratory for studying the process of disk accretion onto magnetic stars ( @xcite and refrences therein ) .
the @xmath0@xmath1 g stellar magnetic field truncates the accretion disk only a few neutron star radii above the surface , channeling the infalling matter toward the magnetic poles .
the resulting x - ray emitting hot spots and accretion shocks are modulated by the stellar rotation to produce the observed pulsations .
such compact accretion geometry poses a particularly interesting challenge .
we want to understand the details of the interaction between the magnetic field and the disk , and the effects that result from the proximity of different emitting regions .
the accretion - powered 2.5 ms pulsar is a particularly good system for this type of study : in addition to having multiple outbursts , it spans orders of magnitude in accretion rate . in this paper
we report our study of the energy - dependence of pulsations from this source and show that it gives new insight into these problems .
was the first detected amp @xcite , and it remains the best studied .
persistent pulsations have been detected throughout the four @xmath01 month outbursts from this source that have been observed by the _ rossi x - ray timing explorer _ ( _ rxte _ ) during 19982005 .
the timing of these pulsations establishes a 2.01 hr binary orbital period @xcite that is increasing on a timescale of @xmath2 yr , possibly due to the ablation of its low - mass companion during x - ray quiescence ( @xcite , hereafter h08 ; @xcite ) .
the presence of x - ray pulsations during the peak accretion rate and the long - term spin - down of this source provide a tight constraint on its magnetic field : @xmath3@xmath4 g ( h08 ; @xcite ) .
a distance of 3.43.6 kpc is estimated from thermonuclear x - ray burst properties @xcite . the outbursts from had similar light curves , which we divide for convenience into four stages ( see fig . 2 of h08 for a schematic ) : a rise from quiescence lasting @xmath55 days ; a short - lived peak at a 225 kev flux of @xmath6@xmath7 erg @xmath8 s@xmath9 ; a 1015 day slow decay in luminosity that lasts until the source reaches @xmath5@xmath10 erg @xmath8 s@xmath9 ; and finally a rapid fall in luminosity followed by weeks of low - luminosity flares with a @xmath05 d periodicity , during which the observed flux can vary by a factor of 1000 @xcite .
this sudden drop has been interpreted as a transition into a `` propeller '' accretion regime ( e.g. , * ? ? ?
* ) in which accretion is only partially inhibited @xcite , although it may also be caused by a change in the viscosity of the accretion disk as it transitions into quiescence @xcite . shortly after the discovery of pulsations in , @xcite noted the presence of soft phase lags , the tendency of the pulses at lower ( `` softer '' ) energies to arrive later than the corresponding pulses at higher ( `` harder '' ) energies .
the magnitude of the lags increased approximately linearly between 210 kev and saturated at 0.08 rotational cycles ( 200 @xmath11s ) above 10 kev .
similar soft lags have been detected in six other amps : igr j00291@xmath125934 @xcite , xte j0929@xmath13314 @xcite , xte j1751@xmath13305 @xcite , xte j1807@xmath13294 @xcite , xte j1814@xmath13338 @xcite , and hete j1900.1@xmath132455 @xcite . a number of explanations for the amp soft lags have been proposed .
models that include only doppler boosting , due to the stellar rotation @xcite or the orbiting material in the accretion disk @xcite , have shortcomings that are summarized by @xcite .
the two most plausible explanations invoke a two - component emission model with differing angular dependence and spectra ( @xcite ; hereafter pg03 ) or the scattering delay of hot photons from the accretion shock off the cooler disk or stellar surface ( @xcite ; hereafter ft07 ) . in this paper
we present an energy - resolved analysis of the persistent pulsations for all the outbursts of detected by _
rxte_. we will discuss the implications of our results for these models of the soft lags .
for our study of the energy dependence of the persistent pulsations , we analyzed the same _ rxte _
observations of as in h08 ( refer to table 1 therein ) .
the ` e_125us_64m_0_1s ` data mode of the _
rxte_proportional counter array ( pca ; @xcite ) was used for nearly all the observations .
it provides 122 @xmath11s time bins and 64 energy channels divided evenly over the detector response , a good combination for discerning the energy dependence of the persistent pulsations . the energy band covered by each channel varied over the 7 yr of observation spanned by this study due to adjustments to the pca anode voltages @xcite
; we accounted for this effect throughout our measurements . for our timing analysis , we shifted the photon arrival times to the solar system barycenter , applied the _ rxte _
fine clock correction , and filtered out data during earth occultations and intervals of unstable pointing .
we also removed any data within 5 min of thermonuclear x - ray bursts . to measure the phases and fractional amplitudes of the persistent pulsations for a given energy band @xmath14 , we folded 512 s intervals of data using the phase timing solution derived in h08 . dividing the resulting pulse profiles into @xmath15 phase bins with photon counts
@xmath16 , @xmath17 , the amplitude ( @xmath18 ) and pulse phase offset ( @xmath19 ) for the @xmath20th harmonicth harmonic is @xmath20 times the frequency of the 401 hz fundamental . ] is @xmath21 we measure the phase offsets of the energy bands relative to the softest band , @xmath22 : @xmath23
. thus negative values of @xmath24 indicate that pulsations of the @xmath20th harmonic within band @xmath14 arrived earlier than the pulsations within the softest band .
the uncertainties of these phase offsets due to poisson noise are @xmath25 where @xmath26 is the total number of photons in the energy band ( h08 ) . for our analysis
, we rejected phase measurements with @xmath27 cycles ( i.e. , 0.1 ms ) .
the measured fractional rms amplitudes are @xmath28 where @xmath29 is the approximate number of background events within our energy band and time interval , estimated using the ftool pcabackest .
uncertainties on the fractional amplitudes are computed using the method described by @xcite and @xcite , which correctly accounts for the addition of noise to the complex amplitude of the signal .
we estimated the 2.525 kev fluxes of the persistent emission by fitting the spectra extracted from standard2 mode data ( 128 energy channels , 16 s time resolution ) with an absorbed blackbody plus power law model , including an iron line if there were significant residuals around 6.4 kev ; see @xcite , 2.1 , for full details .
( relative to the 2.13.3 kev band ) and fractional amplitudes during the slow - decay stages of the 2002 outburst ( black ; mjd 52564.152575.0 ) and the 1998 outburst ( grey ; mjd 50919.050932.8 , the same pointed observations used by @xcite ) .
negative offsets indicate soft lags .
[ fig : avgpulsationsvsenergy],scaledwidth=45.0% ] to assess the energy dependence of the pulse phase and fractional amplitudes , we calculated the average phase offsets and fractional amplitudes during the 11 d slow - decay stage of the 2002 outburst . figure [ fig : avgpulsationsvsenergy ] shows the results , along with measurements from the 1998 outburst for comparison .
we chose this time range because it was the longest and most data - rich interval during which the pulse profile of was observed to be reasonably stable .
the pulse profiles during the slow - decay stage of the 2005 outburst , although similar to the 2002 profiles , were somewhat less stable ( see fig . 3 of h08 ) .
the soft phase lag of the pulsations is immediately apparent for the fundamental .
between 2 kev and 10 kev , the pulses in harder bands arrive progressively sooner , with a slope of @xmath30 cycles kev@xmath9 . between 10 kev and 20 kev
, this trend saturates at a lead of 0.08 cycles ( 200 @xmath11s ) .
there is impressive agreement between the 1998 and 2002 outbursts , which also had nearly identical light curves ; the phase lags during the dimmer 2005 outburst are similar in shape but plateau at a smaller magnitude .
the phase offsets of the second harmonic are less pronounced , disappearing almost entirely or becoming hard lags when the source is bright , as is the case for the data in figure [ fig : avgpulsationsvsenergy ] . in contrast
, the energy dependence of the fractional amplitudes varies considerably between outbursts .
the fractional amplitude of the fundamental during the slow - decay stage of the 1998 outburst ( 5.27.5% rms ) was considerably higher than during the 2002 outburst ( 3.85.0% rms ) , as can be seen in figure [ fig : avgpulsationsvsenergy ] ; the amplitude of the 2005 outburst ( 2.02.6% rms ) was lower still .
the morphology of the energy dependence differs as well .
the 1998 outburst exhibits a large fractional amplitude peak at 3.5 kev and falling amplitudes up to 20 kev , similar to what is seen in xte j1814@xmath13338 @xcite .
in contrast , during 2002 and 2005 the fractional amplitude vs. energy curves are much flatter , with small but significant peaks at 8 kev .
given the strong similarities between the light curves of the 1998 and 2002 outbursts , it is curious that the energy dependence of their fractional amplitudes is markedly different .
all three outbursts show weak evidence of the fundamental s fractional amplitude increasing again above @xmath520 kev .
the fractional amplitude of the second harmonic generally increases with energy for all the outbursts , indicating less sinusoidal pulse profiles in harder energy bands .
xmm observations of the 2008 outburst of this source reveal that the fractional amplitude of both harmonics continues to decrease below 2 kev @xcite .
figure [ fig : bigenergyplot ] illustrates how the energy - dependent phases and fractional amplitudes change over the course of the four observed outbursts .
the soft lags in the fundamental are apparent throughout all four outbursts ; for the second harmonic , soft lags are apparent everywhere except during the outburst peaks . however , the degree of these lags is not constant : they are much more pronounced during the tails of the outbursts , when the flux is low .
@xcite noted a similar pattern in xte j1814@xmath13338 , although that source did not trace as wide a range of fluxes , so the relationship was less clear .
the energy dependence of the fractional amplitudes does not exhibit any clear correlation with the source flux or any other measured parameters , apart from the general trend of fractional amplitudes increasing in the outburst tails ( see 3.4 of h08 ) . to further understand the relationship between the soft lags and the flux , we measured the phase vs. energy slopes over 210 kev .
we chose this range because it has the most significant energy dependence for the fundamental and , to a lesser extent , for the second harmonic .
figure [ fig : phaselagslopevsflux ] plots the resulting slopes as a function of flux . for the fundamental , the magnitude of the soft lags is clearly anticorrelated with the 225 kev flux when the source is bright : as the source dims from its peak at @xmath31 erg @xmath8 s@xmath9 down to roughly @xmath10 erg @xmath8 s@xmath9 , the magnitude of the soft lags grows .
this range of fluxes covers the peak and slow - decay stages of the outburst . at lower fluxes ,
the relation reverses , and the magnitude of the lags becomes correlated with flux .
the flux at which this turnover occurs is significant : once it falls to @xmath10 erg @xmath8 s@xmath9 , the source dims rapidly and enters the flaring - tail stage of the outbursts .
erg @xmath8 s@xmath9 turnover .
these fits are based on the 2002 and 2005 data only , as the 1998 and 2000 outbursts exhibit greater scatter .
[ fig : phaselagslopevsflux],scaledwidth=40.0% ] the trend for the second harmonic is somewhat less clear .
measuring a slope through the 210 kev fractional amplitude points does not work as well since the energy dependence is less linear ( top panel of fig .
[ fig : avgpulsationsvsenergy ] ) , so there is more scatter .
even so , the data strongly suggest a similar relationship at high fluxes , particularly when the relatively scattered 1998 data are excluded .
the slopes of these trends differ significantly for the two harmonics . in the low - flux regime ,
the slopes are @xmath32 for the fundamental and @xmath33 for the second harmonic ; at high fluxes , they are @xmath34 for the fundamental and @xmath35 for the harmonic .
( the units in all cases are @xmath36 cycles kev@xmath9 [ log flux]@xmath9 . )
these slopes were fit using the 2002 and 2005 data only , as the two earlier outbursts had appreciably more scatter and less data .
note that these units reflect _ rotational _ cycles of the star , so a mechanism that induces the same phase lag in all photons of a given energy would result in the slopes being equal for the two harmonics .
for the high - flux slopes , we can conclude with 8@xmath37 significance that this is not the case .
we also looked for correlations between the lags and the spectral parameters from our blackbody plus power law fits .
the only parameter for which a correlation is apparent is the normalization of the hard power law component .
this relation is not surprising , because the hard normalization and the total x - ray flux are linearly proportional : during the 19982005 outbursts , @xmath38 , with a reduced @xmath39 near unity .
the uncertainties of @xmath40 are much greater than the flux uncertainties , particularly when the source is dim , and the large scatter obscures whether a turnover is present in the lag vs. @xmath40 relation .
the energy dependence of the pulsations from exhibit strong variability both between and during the four outbursts observed by _
rxte_. the overall fractional amplitudes and the shape of the amplitude vs. energy curves differ substantially between outbursts , with no obvious correlation with flux ; the only clear trend is that the pulsations generally become less sinusoidal with energy in the observed 230 kev range . in contrast , the soft lags of the pulses have a consistent energy dependence , increasing in magnitude in the 210 kev band and saturating at 1030 kev , with a strong dependence on flux . at the highest fluxes the lags are near zero , but they increase steadily as the flux falls . below a 225 kev flux of @xmath5@xmath41 erg @xmath8 s@xmath9 , this trend reverses and lags get smaller again .
this particular flux also marks the point in the outburst where the luminosity decays rapidly and the source enters the flaring tail .
soft lags are a common feature of amp pulsations , and two plausible models have emerged to explain their existence .
pg03 suggested a two - component model involving soft emission from a surface hot spot and a hard comptonized component from the shocked layer in the accretion funnel .
each component has a different beaming pattern , which is affected in a different way by the doppler effect .
this results in the soft pulse lagging the hard pulse .
while this model works well for and xte j1751@xmath13305 @xcite , it does not predict the reduction in soft lags at very high energies seen in igr j00291@xmath125934 @xcite . as a result ft07 proposed an alternative model in which the disk plays a prominent role . in this model
, the main pulsed element is the hard component from the accretion column : this radiation is then downscattered by cold plasma in the inner disk ( or on the neutron star surface ) to generate a soft lagging component.300 ev in the inner disk @xcite ; the actual temperature will be higher due to the irradiation .
at these temperatures and the expected electron densities ( @xmath42 @xmath43 ; see ft07 and fig . [
fig : falangafitvsmdot ] here ) , all elements through oxygen will easily be fully ionized by the hard tail of the thermal distribution .
the only significant source of photoelectric absorption is fe , but for this @xmath44 only fe xxv and higher remain at @xmath45 ev , with the transition to full ionization occuring at 400600 ev @xcite .
its he - like and h - like forms have k edges at 9 kev , so they will not absorb the softer photons responsible for the lags . since the ft07 model is not ruled out _ a priori _ , this discussion attempts to address both models without preference . ]
these soft lags get larger as the density of the scattering plasma falls .
let us now explore the factors that might cause the flux dependence of the soft lags , assuming that flux directly traces accretion rate .
we then need to ask how the properties of the system are expected to change as the accretion rate falls . for a star with a magnetic field that is strong enough to channel the accretion flow ,
the reduction in accretion rate should cause an increase in inner disk radius ( e.g. , @xcite ) .
this will have two relevant effects .
first , the density of scattering plasma near the star will fall : within the model of ft07 this would cause the soft lags to increase .
in addition the relative angular velocity of the inner disk and star will reduce .
this will affect the angle at which the accretion funnel impacts the surface @xcite , changing the column density above any surface hot spot .
this type of change might work within the model of pg03 to affect the soft lags . if these changes continue steadily
, however , it is hard to see what could cause the turnover seen in figure [ fig : phaselagslopevsflux ] . one intriguing possibility that would give a very natural explanation for such a turnover is that the system reaches the point where the magnetospheric radius @xmath46 exceeds the corotation radius @xmath47 .
this is often referred to as the `` propeller '' regime , since it was originally thought that matter would then be ejected from the system , with the star going from spin - up to spin - down @xcite .
in fact , the transition is probably not sharp , and @xmath46 needs to be substantially larger than @xmath47 for ejection to occur @xcite . at intermediate accretion rates ,
continued accretion is possible even while the neutron star starts to spin down @xcite .
simulations by @xcite indicate that the transition between spin up and spin down should occur when @xmath48 for amps .
the observed turnover happens at an accretion rate of @xmath49 yr@xmath9 , giving @xmath50 for the magnetic field @xmath51 g from h08 . , using @xmath52 kpc and the bolometric correction @xmath53 from @xcite .
the alfvn radius was estimated as @xmath54 , with @xmath55 as per @xcite . @xmath56 and
@xmath57 km were used in all cases . ]
this close agreement between the simulations and the @xmath58 at which the flux rapidly drops and the soft lags turnover strongly suggests that passes through this transition as the accretion rate falls . in this eventuality
there are two effects that could cause the turnover in soft lags . of the inner accretion disk , as fit using the model of ft07 , versus the mass accretion rate for the 2002 and 2005 outbursts of .
the vertical dotted line marks the @xmath58 at which the magnitude of the soft lags is greatest and below which the outburst is in its flaring - tail stage ; a transition from normal accretion to the `` propeller '' regime can explain this change in the observed behavior .
the dashed lines give the best fits of @xmath59 in each region .
[ fig : falangafitvsmdot],scaledwidth=47.0% ] the first is due to the fact that the relative angular velocity of the inner disk and the star changes sign .
the accretion funnel will go from being dragged forwards to being dragged backwards ( assuming that it settles rapidly ; @xcite ) .
the column depth above any surface hot spot should therefore go through a maximum near the zero torque point , causing a turnover in the scattering properties of the accretion column .
if the accretion column plays a major role as discussed in pg03 , this could cause the observed turnover .
another explanation within the pg03 model is the suppression of the lagged soft component at high fluxes due to the increased optical depth of the accretion shock when @xmath58 is large .
however , this mechanism would require the normalization of the soft component to decrease during the outburst peaks , contrary to what is seen ( cf .
fig . 2 of @xcite ) .
a second possibility stems from the fact that the properties of the disk must also change at this point . for accretion to continue after zero - torque point , the inner edge stops retreating , and
the density at the inner edge must rise @xcite . in the scenario of ft07
, this rise in inner disk density would cause the magnitude of the soft lags to decrease again . in the absence of compton upscattering , which would introduce hard lags
not observed in , their model predicts that the scattering from the disk will modify the arrival times as @xmath60 , where @xmath61 is the thompson cross - section and @xmath44 is the electron density of the scattering region .
this model provides reasonable fits to the observed phase lags ; figure [ fig : falangafitvsmdot ] shows the resulting @xmath44 . the fitted electron densities probe the relationship between the inner disk and the accretion rate . @xmath62
@xmath43 gives a mean free path of @xmath63 km and time @xmath64 @xmath11s , requiring @xmath0100 scattering events to produce the observed soft lags .
this @xmath65 rules out the thin neutron star atmosphere as a scattering medium .
on the other hand , simulations by @xcite suggest an inner disk thickness of order the neutron star radius , and @xmath66 is more than adequate to accomodate the number of needed scattering events . from the fits in figure [
fig : falangafitvsmdot ] , we find that @xmath67 during the main outburst ; during the presumed propeller stage , @xmath68 . an interesting consequence of the former relationship is that at higher accretion rates before the turnover , when @xmath69 , the flow rate ( the volume of material per unit time crossing the inner edge of the disk ) must be constant : since @xmath70 , where @xmath71 is the solid angle of the inner edge and @xmath72 is the radial velocity of material crossing it , the flow rate @xmath73 is fixed .
if the phase lags in are entirely due to compton downscattering , as suggested by ft07 , then this model must explain why the magnitudes of the soft lags for the second harmonic are roughly half the magnitudes for the fundamental over a wide range of @xmath58 , as can be seen in figure [ fig : phaselagslopevsflux ] . following the formalism of @xcite , we model the timing effects of comptonization with a distribution @xmath74 of the photon escape times from the scattering medium . for the @xmath20th harmonic of the spin frequency @xmath75 ,
we convolve the harmonic component @xmath76 with @xmath74 to find the new amplitude @xmath77 and the time lag @xmath78 of the scattered pulses : @xmath79 in general , the choice of @xmath74 accounts for the geometry of the emitting and scattering regions .
however , for the large number ( @xmath0100 ) of scatterings previously noted , the geometry becomes unimportant and a photon diffusion distribution of @xmath80 is appropriate @xcite . applying this form to equation ( [ eq : st80formalism ] ) ,
the time lag of the @xmath81 harmonic is @xmath82 the diffusion timescale can not be longer than the lags themselves : @xmath83 @xmath11s .
this sets a lower limit of @xmath84 due to scattering alone , contrary to the observation that @xmath85 .
thus compton scattering can not by itself fully explain the observed lags : a change in the energy dependence of the initial pulses is still necessary .
clearly , detailed calculations are required to check whether the two models really would predict the outcomes that we suggest , but they are at least in principle plausible
. one would also need to test whether the models would predict any reversal in other pulsation properties at propeller onset , given that no such changes are observed .
the interesting behavior of the soft lags in this source suggests that the lags of the other accretion - powered millisecond pulsars should be revisited to check for flux dependence .
the only source for which this has been done is xte j1814@xmath13338 : an increase in soft lags was observed in the tail of the outburst @xcite , but the rapid drop off and shorter tail of this source made it very difficult to study the pulsations at the lowest fluxes .
the other amps may be more promising as candidates for further study .
jmh thanks the max planck institut fr astrophysik for hospitality that facilitated the pursuit of this research .
we would also like to thank lev titarchuk , juri poutanen , and j. martin laming for useful conversations .
this work was supported at mit in part by nasa grants under the rxte guest observer program . ,
k. , swank , j. h. , giles , a. b. , stark , m. j. , strohmayer , t. , zhang , w. , & morgan , e. h. 1996 , in proc .
2808 , p. 59
- 70 , euv , x - ray , and gamma - ray instrumentation for astronomy vii , oswald h. siegmund ; mark a. gummin ; eds . ,
o. h. siegmund & m. a. gummin , 5970 | soft phase lags , in which x - ray pulses in lower energy bands arrive later than pulses in higher energy bands , have been observed in nearly all accretion - powered millisecond pulsars , but their origin remains an open question . in a study of the 2.5 ms accretion - powered pulsar , we report that the magnitude of these lags is strongly dependent on the accretion rate . during the brightest stage of the outbursts from this source , the lags increase in magnitude as the accretion rate drops ; when the outbursts enter their dimmer flaring - tail stage , the relationship reverses .
we evaluate this complex dependence in the context of two theoretical models for the lags , one relying on the scattering of photons by the accretion disk and the other invoking a two - component model for the photon emission . in both cases
, the turnover suggests that we are observing the source transitioning into the `` propeller '' accretion regime . |
the tsc property can be revealed by studying the nature of edge states .
we first calculate the energy dispersion under several conditions . figure [ bands]a shows the evolution of the non - superconducting ( nsc , @xmath26 ) band structure with different values of @xmath14 in brillouin zone ( bz ) corresponding to the 2-site per unit cell . in fig .
[ bands]a , the band crossing below the fermi energy occurs along the @xmath27-@xmath28 direction when there is no existence of the anomalous orbital order ( aoh ) , @xmath29 , and electron and hole pockets are formed at the fermi energy . in the presence of a small value of the aoh term ( @xmath30 ) ,
the band crossing is lifted and the four bands are formed into two disentangled groups . however , the fermi pockets remain .
furthermore , when the topological order is strong enough ( fig .
[ bands]a with @xmath31 ) , an indirect band gap is open directly at the fermi energy and the bulk system becomes insulating , from which no superconducting pairing can emerge .
now , we construct a strip geometry to study the electronic states at the edge with periodic ( fig .
[ bands]b , d ) and open ( fig . [ bands]c , e ) boundary conditions .
since there still exists the translational invariance along the strip direction ( that is , @xmath32-direction ) at the original lattice constant , the momentum along this direction is a good quantum number , which allows to solve the problem for each given wave vector @xmath33 independently .
we can see that the edge bands show up by comparing fig .
[ bands]b , d and c , e . in the metallic / topological - metallic states(@xmath34 ) without the superconducting order , the edge bands have curved features across the fermi surface as highlighted in red in fig .
[ bands]c .
for @xmath29 the non - superconducting ( nsc ) edge states are four - fold degenerate ( including two spin orientations and two edges ) at a given @xmath33 near the region of @xmath35 . when @xmath36 , the degeneracy of edge bands near @xmath35 is reduced to two fold ( that is , only two - fold spin degeneracy ) at a given @xmath33 .
however , these bands are still crossing the fermi energy .
we now turn on the @xmath37-wave pairing ( @xmath38 ) ( see fig .
[ bands]d , e ) . in the absence of the topological order ,
the edge states have no zero - energy modes , which means the quasiparticle states are always gapped . only in the presence of a sufficiently large topological order ,
can the edge states appear to cross the fermi energy ( see fig . [ bands]e with @xmath39 ) .
we note that for @xmath30 , the normal state is metallic and the superconducting pairing is physically feasible .
therefore , we will focus on the case of @xmath30 in the next section . for @xmath31
( see third row of fig .
[ bands]e ) , the normal state is already insulating and the superconducting pairing is not possible .
however , we include this case to show the structure of edge states in response to a `` superconducting '' particle - particle pair . and * b * , @xmath30 .
the high - symmetry momentum points of @xmath28 , @xmath40 and @xmath41 are defined according to the brillouin zone of a 1-site per unit cell .
the superconducting order parameter @xmath38 is taken . ]
the tsc property can also manifest in the local electronic structure around a single impurity .
the use of a single impurity is a powerful method to probe the superconducting pairing symmetry @xcite .
more recently , a universal impurity - induced mid - gap state has also been predicted as a clear signature of topological superfluids @xcite , where the near - zero - energy bound state follows closely follow the symmetry of that of majorana fermions .
inspired by the observation of a robust zero - bias conductance peak in the stm experiments around an interstitial - fe - impurity on the surface of the fe(te , se ) compound @xcite , we construct a microscopic model for the local adsorbed impurity within our model .
figure [ ifi]a shows the 2d lattice structure for a pristine system , where @xmath42-orbital orientation is considered ; while fig .
[ ifi]b shows an adsorbed fe atom and its effective local hopping with the neighboring fe atoms in the parent compound .
we write down the total hamiltonian with the local effect of absorbed site , @xmath43 + { \mathcal{s}}_z^{imp}[\,j\,]\;,\ ] ] in which @xmath44 is the local hopping term from the impurity site to its nearest neighboring fe atoms in the parent system with tunable hopping parameter @xmath45 ( see fig .
[ ifi]b ) , @xmath46 describes the local spin - polarized energy levels on the impurity site with an exchange parameter @xmath47 .
see more details for @xmath44 and @xmath48 in the * methods * section .
we will focus on the local density of states ( ldos ) for fixed topological order @xmath49 and sc pairing strength @xmath50 with varying @xmath45 and @xmath47 . in the regime of a weak hybridization
, one can anticipate that the resulting peaks in the ldos on the adsorbed impurity site merely reflects the energy levels of an isolated atom . in our work
, we are more interested in the regime where the coupling between the adsorbed impurity and its neighbors is relatively strong .
figure [ figldos]a shows the ldos calculated on the adsorbed impurity for various values of @xmath45 and @xmath47 but with fixed @xmath38 and @xmath30 . for comparison , we also show the ldos far away from the impurity site ( black line in fig .
[ figldos]a ) , which resembles the dos in a pristine bulk . for @xmath51
, since the local hopping is still finite , two in - gap peaks appears near the gap edges . with the increased @xmath47 , these in - gap peaks
are shifted toward the fermi energy ( e.g. , in the range of @xmath52 $ ] for @xmath53 as shown in fig .
[ figldos]a ) , and cross the fermi energy ( e.g. , at @xmath54 for @xmath53 as shown in fig . [ figldos]a ) , and move toward the gap edges ( e.g. , in the range of @xmath55 $ ] for @xmath53 as shown in fig . [ figldos]a ) .
we also found a regime of @xmath56 $ ] , for which the in - gap quasiparticle peaks show up again . since in this large @xmath47 regime , the effect of the topological order becomes non - essential , our results are comparable with those in the early study on a magnetic impurity in an iron - based superconductor @xcite
therefore , we do not show here the results for this regime .
the in - gap bound states can also be visualized in the band structure in the reduced brillouin zone of the supercell lattice @xcite . in fig .
[ figldos]b , all the sub - figures are plotted with varying @xmath47 for fixed values of @xmath57 , @xmath30 and @xmath38 .
the localization nature of these in - gap states are investigated by different size of the supercell . as shown in fig .
[ figldos]b , the dispersiveness of the in - gap states is suppressed with increased size of the supercell , suggesting that these in - gap states are truly bound states . to further investigate the consequence of the topological order
, we show in fig .
[ figldos]c the ldos on the impurity site in the absence ( @xmath29 ) and presence ( @xmath30 ) of the topological order .
the zero - energy peak in the ldos in the absence of the topological order is much narrower ( accompanied with higher intensity ) than that in the presence of the topological order , suggesting a much shorter decay length of quasiparticle bound state in the former .
this contrast is further confirmed by the spatial dependence of the ldos at zero - energy , as shown in fig .
[ figldos]d .
specifically , the zero - energy bound state has a longer decay length along the bond direction in a two - dimensional lattice .
this difference can be understood from the band structure of the pristine system in an unfolded brillouin zone , as shown in fig .
[ figbdgband ] , where one can find that the topological order reduces significantly the gap for the wave vector near the @xmath40 point along the @xmath58 path .
we now discuss the relevance of our results to the recent stm experiment on fe(te , se ) compound for an adsorbed fe impurity @xcite . on the one hand ,
the fe(te , se ) compound as mentioned in ref .
has the number of @xmath15-electrons corresponding to the half - filling in the two - orbital model , which is different from the highly electron - doped kfe@xmath0se@xmath0 .
therefore , the chemical potential should be set to ensure the half - filling in our calculations .
in addition , since the distance from the adsorbed fe site to its nearest neighbors are shorter than the fe - fe bonding distance of the parent compound , we anticipate that the local hopping parameter ( @xmath45 ) between the adsorbed fe atom and its neighboring fe atoms is larger than those for the parent compound . on the other hand ,
since the superconductivity for iron - based compounds arises from the fe @xmath59-electrons on an essentially square lattice , the normal - state part of the model hamiltonian originally designed for doped bafe@xmath0as@xmath0 @xcite , to match the low - energy band structure from the lda calculations @xcite , should also be applicable to the fe(te , se ) system .
for the present purpose , we carried out the dft calculations for an fe atom adsorbed in the center of fe plaquette of a large real - space structure of fe(te , se ) compound .
see the * methods * section for the calculation details .
we have found that a spin - polarized fe atom adsorbed on the surface of an otherwise paramagnetic fese compound has a lower energy than a non - spin - polarized adsorbed fe atom .
figure [ figlda]a shows the calculation of ldos for both non - spin - polarized ( upper panel ) and spin - polarized ( lower panel ) adsorbed fe atom .
the broadened peak of the ldos on the adsorbed fe atom itself and the smearing of ldos structure on its neighboring fe sites in the fese compound ( as compared with that on the fe site far way from the adsorbed fe atom ) suggests a strong coupling between the adsorbed fe and its neighboring fe sites .
it supports the treatment of the local hopping parameters in our model calculations . for the case of the spin - polarized case
, the magnetization is mostly localized on the adsorbed fe itself with a proximately induced moment on its nearest neighboring fe sites negligibly small . finally , as shown in fig .
[ figlda]b , each @xmath15-orbital is spin polarized and contributes to the total magnetic moment of 2.89 @xmath12b . however , among the total magnetic moment , only about 14.48% ( about @xmath60 ) arises from each of the @xmath23- and @xmath22-orbitals , which are relevant to the corresponding orbitals active in the pristine compound .
these finding supports the use of a local magnetic impurity with a small spin polarization in our model calculations .
our model study has shown the combined effect of the @xmath7 topological mirror order and the fully gapped @xmath37-wave sc .
this @xmath7 argument extends the mirror symmetry in the mirror - chern representation @xcite for the topological superconductivity of a 2d lattice .
the coexistence of this topological and the superconducting order relies on the intrinsic band structure of the @xmath42 orbitals , which gives a topological metallic state . there
the emergence of gapless edge states in the coexisting phase is a direct evidence for the topological order .
in addition , the introduction of a spin polarized impurity can be regarded as a zero - dimension boundary to the 2d lattice , and it induces a similar quasiparticle behavior as around a vortex core center in the topological superconductor @xcite .
in addition , our model calculations of the local electronic structure have revealed the topological feature of a superconductor through the ldos peak intensity and the size of the impurity - induced in - gap bound states .
finally , our dft calculations of an adsorbed fe atom on the surface of the fese compound has indeed shown the adsorbed impurity is spin polarized , making the findings in the present work relevant to the stm experiment on the fe(te , se ) compound . _ note added . _
as we nearly completed our research , we noticed a recent study by zhang @xcite that described the in - gap bound state around an adsorbed fe atom on the surface of iron - based superconductors , within an anderson impurity model with non - spin polarized @xmath15-electrons on the impurity site . due to the mismatch of energy levels between the impurity and the fe - sites in the background , the qualitative agreement with the stm experiment @xcite
was obtained by adjusting the chemical potential in the pristine system in the strict condition of an unrealistically small hybridization between the adsorbed fe atom and its neighbors .
the shifted chemical potential leads to the system in the highly electron doped regime , which is inconsistent with the fact that fe(se , te ) is in the half - filled regime in the two - orbital model .
in addition , it can be easily understood that with such a small hybridization , the zero - energy peak at the adsorbed fe site has the character of the energy levels of an isolated atom so that the peak intensity is too high .
* model setup . * here
we write down the details of @xmath61+{\mathcal{v}}[\lambda_{aoh}]+{\mathcal{p}}[\delta]$ ] , @xmath62 were @xmath63 are site index , @xmath64 $ ] are orbital index , @xmath65 $ ] is the spin index and @xmath12 is the chemical potential to adjust the fermi surface for hall - filling ( @xmath66 ) . here ,
@xmath67 , are the hopping integrals @xcite . in our previous paper @xcite @xmath14 could be obtained self - consistently from the nn inter - orbital coulomb interaction at the mean - field level . however , in this work , we follow the same spirit as ref . and treat @xmath14 and superconducting pair potential @xmath20 as input parameters for our analysis .
the tensor elements @xmath68 $ ] describe the direction of the nn inter - orbital currents with @xmath69 , and @xmath70 .
the impurity model , @xmath71+{\mathcal{s}}^{imp}_z[j]$ ] , is written , @xmath72 the index , @xmath73 , denotes the impurity site and @xmath74 indicates the nn sites from @xmath73 to its nn sites .
@xmath75 is the local hopping integrals from the ifi site to its nn sites .
based on the d@xmath76 orbital orientation , as shown in fig .
[ ifi]b , there are only two different values in our impurity model @xmath77 ; here we take @xmath78 with @xmath79 . in our calculation ,
different sign and magnitude of @xmath80 do not change our results qualitatively .
the ldos can be calculated according to , @xmath81,\ ] ] where the @xmath82 and @xmath83 are the eigenfunctions of the bdg matrix of the entire hamiltonian .
we use the broadening facor @xmath84 in @xmath85 .
the supercell technique @xcite is used for @xmath27-repeated cell blocks . * dft calculations .
* we use the vienna ab - initio simulation package ( vasp ) @xcite to carry out the lda calculations for an interstitial fe impurity ( ifi ) atom sitting in the center of a large real - space structure of fe(te , se ) compound .
the projector augmented planewave method @xcite and the perdew - burke - ernzerhof @xcite exchange - correlation functional were adopted .
we used a 4@xmath864@xmath862 monkhorst - pack k - point mesh , and a 500 ev cutoff .
the @xmath87-coordinate of fe impurity atoms was fully relaxed up to @xmath88 ev/@xmath89 with other atoms fixed .
we have considered the setup as shown in fig .
[ vasp]a , in which an ifi atom sitting on the top of a two - layered fese . a vacancy space in the z - direction was made to mimic the surface effect of the single ifi atom .
we have calculated(*i * ) non - magnetic(nm ) state and ( * ii * ) ferro - magnetic(fm ) state on the ifi site by turning on and off spin polarization initially on ifi site .
each of the two calculations has reached the charge convergence .
[ vasp]b shows the local distribution of magnetic moment .
the calculated total energy shows that the case ( * ii * ) is more stable than the case ( * i * ) , which confirms a spin polarized ifi site .
we thank s.h.pan for useful discussion in an early stage of our research .
the work at los alamos was supported by the u.s .
doe contract no .
de - ac52 - 06na25396 through the ldrd program ( y .- y.t . &
h.c . ) , the office of basic energy sciences ( bes ) ( t.a . & j .- x.z . ) .
this work was supported in part by the center for integrated nanotechnologies , a doe bes user facility ( j .- x.z . ) .
the work at university of houston was supported in part by the robert a. welch foundation under grant no .
e-1146 and afosr under grant no .
fa9550 - 09 - 1 - 0656 ( c.s.t . ) .
10 yuan - yen tai , c .- c .
joseph wang , matthias j. graf , jian - xin zhu , and c. s. ting , phys . rev .
b * 91 * , 041111(r ) ( 2015 ) .
tai , j .- x .
zhu , m. j. graf , and c. s. ting , europhys .
lett . * 103 * , 67001 ( 2013 ) . j .-
yin , zheng wu , j .- h .
wang , z .- y .
ye , jing gong , x .- y .
hou , lei shan , ang li , x .-
liang , x .- x .
wu , jian li , c .- s .
ting , z. wang , j .-
hu , p .- h .
hor , h. ding , and s. h. pan , arxiv:1403.1027 .
a. y. kitaev , phys .- usp . * 44 * , 131(2001 ) . j. linder , and j. w. a. robinson , nat
. phys . * 11 * , 307(2015 ) .
l. fu , and c. l. kane , phys .
lett . * 100 * , 096407(2008 ) .
i. martin , and a. f. morpurgo , phys .
b * 85 * , 144505 ( 2012 )
. m. m. vazifeh , and m. franz , phys .
lett . * 111 * , 206802 ( 2013 ) .
f. pientka , l. i. glazman , and f. von oppen , phys .
b * 88 * , 155420 ( 2013 ) .
j. klinovaja , p. stano , a. yazdani , and d. loss , phys .
lett . * 111 * , 186805 ( 2013 ) .
i. reis , d. j. j. marchand , and m. franz , phys . rev .
b * 90 * , 085124 ( 2014 ) .
s. nadj - perge , i. k. drozdov , j. li , h. chen , s. jeon , j. seo , a. h. macdonald , b. a. bernevig , a. yazdani , science * 346 * , 602 ( 2014 ) .
qi , taylor l. hughes , and shou - cheng zhang , phys .
b * 82 * , 184516(2010 ) . fan zhang , c. l. kane , and e. j. mele , phys .
lett . * 111 * , 056403(2013 ) .
qi , taylor l. hughes , s. raghu , and shou - cheng zhang , phys . rev .
lett . * 102 * , 187001(2009 ) .
yuji ueno , ai yamakage , yukio tanaka , and masatoshi sato , phys . rev . lett . * 111 * , 087002(2013 ) .
h. chen , y .- y .
tai , c. s. ting , m. j. graf , j. dai , and j .- x .
zhu , phys .
b * 88 * , 184509 ( 2013 ) .
a. v. balatsky , i. vekhter , and j .- x .
zhu , rev .
* xxx * , xxx ( 2006 ) .
xia - ji liu , and hui hu , phys .
a * 85 * , 033622 ( 2012 ) .
hui hu , lei jiang , han pu , yan chen , and xia - ji liu , phys .
lett . * 110 * , 020401 ( 2013 ) .
jian - xin zhu , b. friedman , and c.s .
ting , phys .
b * 59 * , 3353 ( 1999 ) .
wei - feng tsai , yan - yang zhang , chen fang , and jiangping hu , phys .
b * 80 * , 064513 ( 2009 ) .
alaska subedi , lijun zhang , d. j. singh , and m. h. du , phys .
b * 78 * , 134514 ( 2008 )
. masatoshi sato , and satoshi fujimoto , phys .
b * 79 * , 094504 ( 2009 ) .
degang zhang , arxiv:1408.0567 ( 2014 ) .
jian - xin zhu , i. martin and a. r. bishop , phys .
* 100 * , 236403 ( 2008 ) .
g. kresse , d. joubert , `` from ultrasoft pseudopotentials to the projector augmented - wave method '' , phys .
b 59 , 1758 ( 1999 ) .
g. kresse , j. furthmller , `` efficient iterative schemes for ab initio total - energy calculations using a plane - wave basis set '' , phys . rev .
b 54 , 11169 ( 1996 ) .
j. p. perdew , k. burke , and m. ernzerhof , generalized gradient approximation made simple " , phys . rev .
lett , 77 , 3865 ( 1996 ) . | recently topological superconducting states has attracted a lot of interest . in this work ,
we consider a topological superconductor with z@xmath0 topological mirror order @xcite and s@xmath1-wave superconducting pairing symmetry , within a two - orbital model originally designed for iron - based superconductivity @xcite .
we predict the existence of gapless edge states .
we also study the local electronic structure around an adsorbed interstitial magnetic impurity in the system , and find the existence of low - energy in - gap bound states even with a weak spin polarization on the impurity .
we also discuss the relevance of our results to the recent stm experiment on fe(te , se ) compound with adsorbed fe impurity @xcite , for which our density functional calculations show the fe impurity is spin polarized .
topological superconductor ( tsc ) has been attracted lots of interest on its potential application to the fault - tolerance quantum computation @xcite and superconducting spintronics @xcite .
the ts can be designed through the interface of a topological insulator with the fully gapped superconductor @xcite or by engineering through the interface of ferromagnetic - superconducting nanostructure under several special conditions @xcite .
most of these efforts were aimed to look for the _ edge states _ and _ in - gap bound states _ in a fully gapped superconductor .
there are two types of edge states for a two - dimensional ( 2d ) tsc .
the first type is a _
chiral _ edge state in a time - reversal - symmetry - broken superconducting pairing state like @xmath2_{\uparrow\uparrow}$ ] @xcite .
the second type is a _ helical _ edge state in a time - reversal - symmetry - invariant pairing state like @xmath3_{\uparrow\downarrow}$ ] ) @xcite .
recently , the crystalline symmetry protected topological phase with mirror chern number has been studied in a triplet pairing state of sr@xmath0ruo@xmath4 @xcite , for which chiral edge states are obtained @xcite .
a realization of helical edge state can be established in a junction formed by an @xmath5-wave superconductor and a strong topological insulator ( ti ) , mimicking a time - reversal - symmetry - preserved @xmath6-wave superconducting pairing state @xcite . in this work ,
we propose a feasible scenario to realize a helical topological mirror superconductor ( htms ) based on the coexistence of the interaction - driven @xmath7 topological mirror order and @xmath8-wave pairing symmetry .
our concept is based on the nature of a topological metal , where the superconductivity can be generated through a finite fermi surface with electron-(hole- ) fermi pockets .
the @xmath7 mirror topological order creates robust edge states even in the presence of a fully gapped @xmath8-wave pairing symmetry .
we further analyze the local electronic structure of an adsorbed interstitial magnetic impurity ( imi ) in the 2d bulk of the htms , and identify the different consequence from the topological and non - topological superconductor .
it is important to note that , since the superconductivity is intrinsically inherited from the normal - state band structure , our proposal avoids the need to make a ti - sc heterostructure for helical edge states . to make our model into a real - material context
, we also carry out the dft calculations for an adsorbed interstitial - fe impurity in recently discovered 11-family of iron - based superconductors fese , and find the adsorbed fe - atom to be spin polarized . therefore , our results have a direct relevance to the recent stm experiment on fe(te , se ) scs , where a robust zero - energy bound state ( zbs ) has been found around an adsorbed fe atom @xcite . we consider the following effective hamiltonian to describe the coexistence of the @xmath7 topological mirror order with the @xmath8-pairing symmetry , @xmath9+{\mathcal{v}}\,[\,\lambda_{aoh}\,]+{\mathcal{p}}\,[\,\delta\,].\ ] ] here @xmath10 is the kinetic energy with optimized hopping terms ( @xmath11 ) and chemical potential ( @xmath12 ) to describe the low energy physics that originally designed for the fe - based compound @xcite
, @xmath13 is responsible for the emergent @xmath7 topological mirror order @xcite with tunable anomalous orbital hall order ( @xmath14 ) for a @xmath15-wave form factor @xmath16_{\alpha,\bar \alpha}$ ] ( where @xmath17 , @xmath18 ) , and @xmath19 is the @xmath8-wave pairing order ( @xmath20 ) with the form factor @xmath21_{\uparrow\downarrow}$ ] ( in 1-fe bz ) .
see definitions of @xmath10 , @xmath13 and @xmath19 in the * method * section . throughout the work
, the chemical potential is adjusted to ensure the system is half filled . in the following section , we numerically solve the band structure of eq .
( [ model ] ) to reveal the interplay of topological ( @xmath14 ) and superconducting ( @xmath20 ) orders . and
@xmath22 orbitals .
* b * , the position of an interstitial atom located at the center of a plaquette of the 2d square lattice together with the local hopping process among the @xmath23/@xmath22 orbitals , where @xmath24 stand for orbital lobes aligned parallel ( perpendicular ) to the hopping direction between two nearest - neighboring sites of the square lattice ) .
we set @xmath25 for the parameterization . ] |
helical magnets are systems in which long - range magnetic order takes the form of a helix or spiral , such that in any given plane perpendicular to a preferred direction there is ferromagnetic order , but the direction of the magnetization rotates as one goes along the preferred axis .
the pitch vector @xmath0 of the helix points in the preferred direction , and its modulus @xmath1 is the pitch wave number , with @xmath2 equal to the helical wavelength .
one mechanism for stabilizing this type of order over the homogeneous ferromagnet was pointed out by dzyaloshinksky@xcite and moriya.@xcite it relies on the spin - orbit interaction , which can lead , for certain lattice structures , to a term of the form @xmath3 in the hamiltonian or action , with @xmath4 the magnetic order parameter .
the presence of such a chiral term implies that a homogeneously magnetized state can always gain energy by means of a nonzero curl of the magnetization , which leads to a helical ground state . in a rotationally invariant system
the direction of the pitch vector would be arbitrary ( analogous to the arbitrary direction of the magnetization in an isotropic ferromagnet ) .
however , the underlying lattice leads , via crystal - field effects , to a pinning of the helix in certain directions determined by the lattice .
the crystal - field effects are due to the spin - orbit interaction , as is the dzyaloshinsky - moriya ( dm ) mechanism for helical order itself .
since the spin - orbit interaction is weak on an atomic or microscopic scale , with a dimensionless coupling constant @xmath5 , this leads to a hierarchy of energy or length scales that can be classified according to their dependence on powers of @xmath6 .
finally , an external magnetic field provides an additional energy scale , and couples to the helix via an induced homogeneous component of the magnetization .
this leads to a rich phase diagram that is the topic of the present paper .
a very well - studied helimagnet of dm - type is the metallic system mnsi , which we will concentrate on ; a very similar , but less extensively studied material is fege.@xcite in zero magnetic field , and at ambient pressure , mnsi displays long - ranged helical order with @xmath7 below a temperature @xmath8.@xcite mnsi crystallizes in the cubic b20 structure , and the helix is observed to be pinned in the @xmath9-directions.@xcite the corresponding space group is p2@xmath103.@xcite in a magnetic field @xmath11 , the helix is superimposed by a homogeneous component of the magnetization , which leads to the so - called conical phase.@xcite a magnetic field in the @xmath12$]-direction tilts the helix away from @xmath13 until the @xmath0-vector aligns with the magnetic field at a critical field strength @xmath14 .
with increasing @xmath11 the amplitude of the helix decreases , and finally vanishes at a critical field @xmath15 , where the system enters a field - polarized ferromagnetic phase . inside the conical phase at intermediate fields near @xmath16
there is a region known as the a - phase.@xcite this was thought to represent a helix whose pitch vector is perpendicular to the magnetic field,@xcite but recently has been interpreted as a topological phase where three helices with co - planar @xmath17-vectors form a skyrmion - like structure.@xcite the schematic phase diagram in the @xmath11-@xmath18-plane as observed experimentally is shown in fig .
[ fig:1 ] .
in addition to the phases shown , a possible second transition and spin - liquid phase at @xmath19 just above @xmath16 has been reported recently.@xcite -0 mm -@xmath18 plane showing the helical , conical , and a - phases , as well as the field - polarized ferromagnetic ( fm ) and paramagnetic ( pm ) states . see the text for a description of these phases.,title="fig:",width=226 ] another interesting aspect of mnsi is its sensitivity to hydrostatic pressure .
with increasing pressure @xmath20 , the magnetic transition temperature @xmath16 decreases until it vanishes at @xmath21.@xcite the transition is second order or very weakly first order above a temperature of approximately @xmath22 , and strongly first order at lower temperatures , with a tricritical point separating the two regimes .
these features have been explained as universal properties of quantum ferromagnets in an approximation that neglects the helical order at longer length scales.@xcite in the paramagnetic region at pressures up to approximately @xmath23 and at temperatures @xmath24 , strong non - fermi - liquid transport properties are observed , with the electrical resistivity @xmath25 displaying a temperature dependence @xmath26 over almost three decades in temperature.@xcite the origin of this behavior has recently been proposed to be a combination of columnar spin textures and very weak quenched disorder.@xcite in a smaller region , above @xmath27 and below a pressure - dependent temperature @xmath28 , short - range ( spin - liquid ) helical order has been observed@xcite and various explanations in terms of analogs of blue phases in liquid crystals have been proposed.@xcite recently , an anomalous hall effect has been observed at intermediate pressures and below @xmath16 in the helically ordered phase , and it has been suggested that this feature is related to the short - ranged order observed above @xmath27.@xcite the phase diagram in the @xmath18-@xmath20 plane is shown in fig .
[ fig:2 ] .
-0 mm -@xmath20 plane .
the tricritical point ( tcp ) separates a line of second order transitions ( solid line ) from a line of first - order transitions ( dashed line ) .
the inset shows the resistivity data from ref . in the non - fermi - liquid ( nfl ) region .
the boundary of the nfl region ( dotted line ) is not sharp .
sl1 , sl2 , and sl3 refer to the possible spin - liquid phases or regions reported in refs . and , , and , respectively.,title="fig:",width=264 ]
if a magnetic field is applied in the vicinity of @xmath29 , tricritical wings , i.e. , surfaces of first - order transitions , emerge from the tricritical point that are believed to end in a pair of quantum critical points in the @xmath30 plane.@xcite this feature , which is depicted in fig .
[ fig:3 ] , has been explained theoretically in ref .. -0 mm , @xmath20 , and @xmath11 showing the tricritical wings and quantum critical points ( qcp).,title="fig:",width=302 ] if one ignores crystal - field effects , i.e. , in an isotropic model , the long - range helical order in the helical phase ( @xmath19 , @xmath31 , @xmath32 ) breaks the translational symmetry .
this spontaneous breaking of a continuous symmetry leads to a goldstone mode , the helimagnon.@xcite the dispersion relation is anisotropic , and given by _
( k ) = . [ eq:1.1 ] here the wave vector @xmath33 has been decomposed into components parallel and perpendicular , respectively , to the helical pitch vector @xmath0 , and @xmath34 and @xmath35 are elastic constants .
the absence of a term proportional to @xmath36 under the square root in eq .
( [ eq:1.1 ] ) reflects rotational symmetry and is analogous to the nature of the goldstone mode in either smectic or cholesteric liquid crystals.@xcite crystal - field effects that couple the electrons to the underlying lattice break the rotational invariance and lead to a @xmath36 term with a small prefactor , i.e. , the dispersion relation becomes phonon - like .
both the single - particle relaxation rate @xmath37 that results from the scattering of electronic quasi - particles by helimagnons and the thermal resistivity @xmath38 have a non - fermi - liquid temperature dependence @xmath39 .
in contrast , the transport relaxation rate that determines the electrical conductivity @xmath25 in a boltzmann approximation has a temperature dependence @xmath40.@xcite the properties of the helical phase in the presence of weak quenched disorder have also been investigated.@xcite in the present paper we will discuss the derivation of the phase diagram sketched above , as well as some refinements , within the context of a mean - field theory and with an emphasis on the @xmath11-@xmath18 plane .
we will also determine the nature of the goldstone modes , and their consequences for electronic properties , in various observed or proposed phases with long - range order .
in addition to the specific heat , the single - particle relaxation rate , and the electrical conductivity we will consider the thermal resistivity @xmath38 , whose temperature dependence is the same as that of the single - particle relaxation rate . we will consider a landau - ginzburg - wilson ( lgw ) functional for a three - dimensional order parameter ( op ) field @xmath41 whose expectation value is proportional to the magnetization .
we will consider an action that is appropriate for the helical magnets mnsi and fege , which crystallize in the cubic b20 structure with space group p2@xmath103 .
we will organize the action according to the dependence of its various constituents on powers of the spin - orbit interaction @xmath6 . within this scheme ,
we write = a _ + a _ + a _ , [ eq:1.2 ] where _ & = & _ v dx [ eq:1.3 ] is the action for an isotropic classical heisenberg ferromagnet in a homogeneous external magnetic field @xmath42 .
@xmath43 denotes a real - space integral over the system volume .
@xmath44 stands for @xmath45 , with @xmath46 the components of the gradient operator @xmath47 .
@xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 are the parameters of the landau theory and are related to the microscopic energy and length scales ( see sec.[subsubsec : i.b.2 ] below ) ; they are of zeroth order in the spin - orbit coupling @xmath6 .
equation ( [ eq:1.3 ] ) contains all analytic terms invariant under co - rotations of real space and op space up to quartic order in @xmath4 and bi - quadratic order in @xmath4 and @xmath53.@xcite we have added one higher - order term , with coupling constant @xmath52 , as an example of a class of terms that can stabilize unusual phases in helimagnets , although they are not of qualitative importance in ferromagnets .
we will consider this term only in appendix [ app : a ] .
[ eqs:1.4 ] _
= _ v dx m(x)((x ) ) [ eq:1.4a ] is the chiral dzyaloshinksy - moriya term that favors a nonvanishing curl of the magnetization .
the existence of this term hinges on the spin - orbit coupling , as well as on the system not being invariant with respect to spatial inversion ( due to the linear dependence on the gradient operator ) .
the coupling constant @xmath54 is linear in @xmath6 , and on dimensional grounds we have c = a , [ eq:1.4b ] with @xmath55 the fermi wave number which serves as the microscopic inverse length scale . in this context , this can be considered the definition of @xmath6 .
the preceding contributions to the action are all invariant under either separate rotations , or co - rotations , in spin ( or magnetization ) space and real space . the spin - orbit interaction couples the electron spin , and hence the magnetization , to the underlying lattice .
therefore , in addition to the rotationally invariant terms , any term that is invariant under elements of the space group connected with the crystal lattice is allowed . for the b20 structure of mnsi and fege ,
the appropriate space group is p2@xmath103 . to quartic order in @xmath4 , and bi - quadratic order in @xmath53 and @xmath4 , the allowed terms in the action are the crystal - field terms [ eqs:1.5 ] _ & = & _ v dx _ i=1 ^ 3 [ eq:1.5a ] where @xmath56 .
the last term is the usual cubic anisotropy that is always present in a magnet on a cubic lattice , and v = u ^4 [ eq:1.5b ] with @xmath57 . of the gradient - squared terms ,
the first one also has cubic symmetry ; the second one does not , but is invariant under elements of p2@xmath103 . on dimensional grounds
, we have b & = & a ^2 , + b_1 & = & a_1 ^2 , [ eq:1.5c ] with @xmath58 .
the various contributions to the action @xmath59 , and their dependencies on @xmath6 , imply a hierarchy of energy scales and corresponding wave number or length scales . at zeroth order in @xmath6
, we have the microscopic scale , which is represented by the fermi energy @xmath60 and the fermi wave number @xmath55 .
fluctuations renormalize this to the critical scale , which is represented by the magnetic ordering temperature @xmath16 and the corresponding length scale .
the physics at these scales is described by @xmath61 , eq.([eq:1.3 ] ) .
the chiral dm term is balanced by the rotationally invariant gradient squared term in eq .
( [ eq:1.3 ] ) that makes magnetization gradients energetically costly . as a result , the relevant gradient or momentum scale is of @xmath62 , and hence the chiral wave number scale is given by the microscopic scale times @xmath6 .
this determines the parameters of the helix , in particular the helical pitch wave number @xmath63 . in mnsi and fege , this wave number scale is on the order of 100 times smaller than the microscopic scale .
@xmath64 contains one explicit factor of @xmath6 and one gradient , and hence its contribution to the free energy is of @xmath65 .
the physics at this scale is described by @xmath66 in conjunction with @xmath67 . at fourth order in @xmath6 ,
crystal - field effects come into play .
they pin the helix , are small compared to the chiral energy scale by another factor of @xmath68 , and are described by @xmath69 , eqs .
( [ eqs:1.5 ] ) .
since gradients are effectively of @xmath62 , see above , the contributions of all three terms in @xmath69 to the free energy are of @xmath70 . finally , the external magnetic field sets a scale that is continuously tunable .
we now derive the mean - field phase diagram for systems described by the action given in eqs .
( [ eq:1.2 ] ) - ( [ eqs:1.5 ] ) .
we will use the hierarchy of energy scales explained in sec .
[ subsec : i.b ] to show how a more and more sophisticated phase diagram emerges as one keeps effects of higher and higher order in @xmath6 .
to do so , we consider field configurations of the following form : [ eqs:2.1 ] ( x ) = m_0 + m_1^+_+ ( q ) + m_1 ^ -_-(q ) .
[ eq:2.1a ] here @xmath71 is a homogeneous component of the magnetization , @xmath72 are amplitudes of fourier components with wave vector @xmath0 , and @xmath73 are two unit vectors that form a right - handed _
dreibein _ together with @xmath0 : _ + _ - = , _ + = _ - , _ - =
_ + , [ eq:2.1b ] where @xmath74 .
the sinusoidal terms in eq .
( [ eq:2.1a ] ) describe a helix with pitch vector @xmath0 .
the helix is in general elliptically polarized , and it is useful to define a polarization parameter = m_1 ^ -/m_1^+ , [ eq:2.2 ] special cases are circular polarization , @xmath75 , and linear polarization , @xmath76 or @xmath77 . the motivation for the _ ansatz _ eqs.([eqs:2.1 ] ) is provided by the fact that it gives the functional form of the global minimum of the action @xmath59 , eqs .
( [ eq:1.2 ] ) - ( [ eqs:1.5 ] ) , if one neglects the crystal - field terms @xmath78 ; that is , for the action up to @xmath65 , see ref . and sec .
[ subsec : ii.b ] below .
to zeroth order in @xmath6 the system is approximated by a ferromagnet . according to the action @xmath61 , eq . ( [ eq:1.3 ] ) , for @xmath19 there is a second - order phase transition which in mean - field approximation occurs at @xmath80 .
for @xmath81 there is a crossover at @xmath80 from a field - polarized paramagnetic state , where the magnetization extrapolates to zero for @xmath82 , to a field - polarized ferromagnetic state where the magnetization extrapolates to @xmath83 . the free energy density in mean - field approximation
, @xmath84 , in a zero field is f = -t^2/4u .
[ eq:2.3 ] in a nonzero field , one has [ eqs:2.4 ] f = m_0 ^ 2 + m_0 ^ 4 - hm_0 , [ eq:2.4a ] where @xmath85 is the solution of the mean - field equation of state tm_0 + um_0 ^ 3 = h. [ eq:2.4b ] this is just a classical heisenberg model , so it can not explain the tricritical point and the associated tricritical wings .
the latter emerge within a renormalized mean - field theory that takes into account the coupling of the magnetization to other electronic degrees of freedom.@xcite this leads to a fluctuation - induced first - order transition in analogy to the case of the nematic - to - smectic - a transition in liquid crystals,@xcite as well as to the tricritical wings and the quantum - critical points in an external magnetic field.@xcite the ferromagnetic approximation suffices for understanding the gross features of the phase diagram in the @xmath18-@xmath20 plane .
to second order in @xmath6 we need to add the dm term , eq .
( [ eq:1.4a ] ) , to the action .
it is obvious that this term favors a nonzero curl of the magnetization , with the direction of the curl depending on the sign of @xmath54 .
however , the spatial variation of @xmath4 will be limited by the other gradient terms in the action , the @xmath44 term in particular .
we thus expect a spatial modulation of @xmath4 on a length scale on the order of @xmath86 .
it is easy to check that the ansatz , eqs .
( [ eqs:2.1 ] ) , with @xmath87 , i.e. , @xmath88 , and [ eqs:2.5 ] & = & q , [ eq:2.5a ] + m_0 & = & m_0 , [ eq:2.5b ] + m_0 & = & h/(cq - aq^2 ) , [ eq:2.5c ] + m_1 ^ 2 & = & ( -t + cq -aq^2 - ) , [ eq:2.5d ] solves the saddle - point equations for the action @xmath89 . in order to determine @xmath90
, we extremize the resulting free energy with respect to @xmath90 .
@xmath91 is a solution for all values of @xmath11 . finally , we need to ascertain that the solution is a minimum , which turns out to be the case for @xmath92 and @xmath93 .
we thus find that the field configuration [ eqs:2.6 ] ( x ) = m_0 + m_1 , [ eq:2.6a ] with @xmath94 , @xmath95 , @xmath96 forming a _ dreibein _ , and q & = & c/2a , [ eq:2.6b ] + m_0 & = & h / aq^2 , [ eq:2.6c ] + m_1 & = & , [ eq:2.6d ] where [ eqs:2.7 ] r & = & t - aq^2 , [ eq:2.7a ] + & = & aq^2 , [ eq:2.7b ] minimizes the free energy in the parameter range @xmath97 , @xmath98 .
we will refer to this state as the aligned conical state ( acs ) , to distinguish it from the perpendicular conical state discussed in appendix [ app : a ] .
the acs is actually a global minimum , as can be seen by writing the action as a sum of positive semi - definite terms that are individually minimized by this state.@xcite the mean - field free energy density in that range is f = ( + ) .
[ eq:2.8 ] equations ( [ eqs:2.6 ] ) describe the helical phase for @xmath19 and the conical phase for @xmath99 . comparing eqs .
( [ eq:2.3 ] ) and ( [ eq:2.8 ] ) we see that the helical transition preempts the ferromagnetic one
. for @xmath100 from below the helical component of the magnetization vanishes , and the free energy , eq .
( [ eq:2.8 ] ) , approaches that of the ferromagnet , eqs .
( [ eqs:2.4 ] ) . for @xmath101
the equation of state and the free energy for the dm action @xmath102 are the same as for a ferromagnet with action @xmath61 .
these considerations account for the structure of the phase diagram shown in fig .
[ fig:1 ] except for the field @xmath14 and the a - phase .
recent neutron - scattering experiments by mhlbauer et al.@xcite are consistent with the notion that the a - phase is characterized by spin textures that form line defects in the direction of the magnetic field , with the lines forming a hexagonal lattice , the a - crystal , in the plane perpendicular to the field .
this experimental observation led the authors of ref . to suggest a skyrmion state consisting of three co - planar helices as underlying the a - phase : ( x ) & = & m_0 + & & -60pt + m_1 _
i=1 ^ 3 , [ eq:2.9 ] where the pitch vectors @xmath103 all have the same modulus @xmath90 , are perpendicular to @xmath42 , the angle between adjacent pitch vectors is @xmath104 , and @xmath105 , @xmath106 , and @xmath107 form a right - handed _
dreibein _ for each value of @xmath108 .
the values of @xmath90 , @xmath85 , and @xmath109 are obtained by minimizing the action .
the free energy difference between this state and the conical one has a minimum at @xmath110 , but it is still positive even at the minimum .
however , ref . found that taking into account gaussian fluctuations stabilizes the state with respect to the conical one .
it should be noted that eq .
( [ eq:2.9 ] ) is not a solution of the saddle - point equations for the action given by eqs .
( [ eq:1.2])- ( [ eqs:1.5 ] ) , and therefore can not be a true local minimum of the free energy . also , the relation between this _ ansatz _ and what are commonly called skyrmionic spin configurations , which _ are _ solutions of the saddle - point equations ( see ref . and references therein ) , is not clear .
most likely it represents a single - fourier - component approximation to a saddle point and is a bloch - type description of skyrmions as opposed to a wannier - type description of isolated skyrmions in ref .. in any case , it describes a hexagonal array of line defects with the spin antiparallel to the magnetic field at the defect centers , and parallel at points on the cell boundaries , see fig .
[ fig:4 ] , in qualitative agreement with the neutron scattering data .
-0 mm ) with the @xmath0 vectors in the @xmath111-@xmath112 plane , @xmath113 , @xmath114 , and @xmath115 .
the arrows represent projections of the spins into the plane .
the dark and light regions denote spin directions antiparallel and parallel to @xmath116 , respectively.,title="fig:",width=302 ] earlier , grigoriev et al.@xcite had proposed a single - helix state with the pitch vector oriented perpendicular to the external field .
although current experimental evidence favors a skyrmion state as a candidate for the a - phase , it is still of interest to discuss such a perpendicular conical state ( pcs ) , since it might be a viable candidate for the ground state in some other part of the phase diagram of helimagnets .
we therefore briefly discuss the pcs and its properties in appendix [ app : a ] .
the skyrmion lattice approximately described by eq .
( [ eq:2.9 ] ) can melt , which will lead to a skyrmion liquid .
in such a state the line defects shown in fig . [ fig:4 ] still exist , but they no longer form a lattice .
rather , their fluctuations in the plane perpendicular to the line , which are illustrated in fig .
[ fig:5 ] , have become so large that the long - range order is destroyed .
such a state has recently been proposed to represent the nfl region shown in fig .
[ fig:2].@xcite although it is not an ordered phase , such a state has much in common with the a - phase and we will discuss it in the context of the ordered phases .
-0 mm and of fluctuations about this state .
@xmath49 is the lattice constant , and @xmath117 is a displacement vector.,title="fig:",width=188 ] to fourth order in @xmath6 , we need to take into account the crystal - field terms shown in eq .
( [ eq:1.5a ] ) .
this makes the saddle - point equations very complicated , and no exact solution is known .
we therefore take a variational approach by inserting eq .
( [ eq:2.1a ] ) into the action and minimizing with respect to the parameters of the _ ansatz_. of all the members of the class of functions represented by eqs .
( [ eqs:2.1 ] ) this will yield the one with the lowest free energy . by writing eq .
( [ eq:1.5a ] ) we have fixed the coordinate system by choosing the crystallographic axes to be the @xmath118 axes .
we thus are no longer free to choose the direction of @xmath42 , @xmath119 , or @xmath0 . for simplicity
, we will consider only the case of a magnetic field along the @xmath120-axis : @xmath121 .
@xmath122 we parameterize in terms of angles @xmath123 and @xmath124 as follows , [ eqs:2.10 ] = ( , , ) ( _ 1 , _ 2 , _ 3 ) , [ eq:2.10a ] with @xmath125 .
this leaves one free parameter for @xmath126 , namely , an azimuthal angle @xmath127 : _ + & = & ( , _ e - _ e , + & & _ e + _ e , + & & -_e ) .
[ eq:2.10b ] this uniquely determines @xmath128 .
finally , @xmath71 in general needs to be decomposed into components parallel and perpendicular , respectively , to @xmath17.@xcite however , while a perpendicular component can lead to a slightly lower free energy , it has no qualitative effects on the structure of the phase diagram , and we therefore restrict our ansatz to _ 0 = m_0 .
[ eq:2.10c ] we further assume that the system is sufficiently close to a second order or weakly first order phase transition that one can neglect the last term in eq.([eq:1.5a ] ) . with these approximations ,
the free energy does not depend on the angle @xmath127 and is completely parameterized in terms of six parameters , namely : two amplitudes , @xmath85 and @xmath129^{1/2}/\sqrt{2}$ ] , the polarization parameter @xmath130 , the modulus @xmath90 , and the two direction angles @xmath123 and @xmath124 .
we find [ eqs:2.11 ] f & = & tm_0 ^ 2 + ( m_0 ^ 2 + m_1 ^ 2 ) + ( m_0 ^ 2 + m_1 ^ 2)^2 + m_1 ^ 2- h m_0 + & & + m_1 ^ 2q^2b _ ( , ) + m_1 ^ 2q^2b_1 ( , ) - m_1 ^ 2q^2b _ ( , ) - m_1 ^ 2q^2b_1 ( , ) + o(^8 ) .
[ eq:2.11a ] here we have defined @xmath131 , with @xmath132 arbitrary at this point . we also have made use of the fact that we know , from sec.[subsec : ii.b ] , that the physical solution has the property @xmath133 with @xmath134 , and have expanded in powers of @xmath135 .
the angle - dependent functions in eq .
( [ eq:2.11a ] ) are b _ ( , ) & = & b^+ ( , ) b^- ( , ) , + b_1 ( , ) & = & b_1^+ ( , ) b_1 ^ - ( , ) , [ eq:2.11b ] where b^+ ( , ) & = & 2 ^2 ^2 ^2 , + b^- ( , ) & = & ^2 ^2(1 + ^4 + ^4 ) , + b_1^+ ( , ) & = & ^2 ^4 + ^2 ^2 , + b_1 ^ - ( , ) & = & ^4 ^2 + ^4 ^2 + & & + ^2 ^ 2 ^2 ^2 . [ eq:2.11c ]
we now need to minimize the free energy with respect to the six parameters .
we will first consider the case @xmath19 to understand the pinning of the helix by the crystal - field terms , and then determined the effects of a magnetic field .
furthermore , in order to keep the discussion transparent we will initially restrict ourselves to an _ ansatz _ with a circular polarization , @xmath136 .
this suffices to understand the existence of the critical field @xmath14 .
we will then generalize the _
ansatz _ to allow for a non - circular polarization and show that this leads to a splitting of the transition at @xmath14 , with a first order transition from a circularly polarized helix to an elliptically polarized one at a critical field @xmath137 preceding the alignment transition at @xmath14 .
we consider the system at @xmath19 and initially restrict our _ ansatz _ to the case of circular polarization , @xmath138 . the remaining angular dependence in the free energy , eq .
( [ eq:2.11a ] ) , is contained in the functions @xmath139 and @xmath140 . minimizing with respect to @xmath124
we find that there are two cases .
_ case ( 1 ) : _ : : @xmath141 .
this implies @xmath142 or @xmath143 , and @xmath144 , @xmath145 .
minimizing with respect to @xmath123 leads to two subcases : + _ case ( 1a ) : _ ; ; @xmath146 .
the free energy is minimized by @xmath147 , which implies @xmath148 or equivalent .
_ case ( 1b ) : _ ; ; @xmath149 .
the free energy is minimized by @xmath150 , which implies @xmath151 or equivalent .
_ case ( 2 ) : _ : : @xmath152 .
this implies @xmath153 or equivalent , and @xmath154 , @xmath155 . minimizing with respect to @xmath123 yields + _ case ( 2a ) : _ ; ; @xmath146 .
the free energy is minimized by @xmath156 , which implies @xmath157 or equivalent .
_ case ( 2b ) : _ ; ; @xmath149 .
the free energy is minimized by @xmath158 , which implies @xmath159 . by comparing the resulting free energies for these cases we see that case ( 1b )
provides the minimum for @xmath160 , whereas case ( 2a ) provides the minimum for @xmath161 .
this is a generalization of the result obtained in ref . , which considered a model with @xmath162 .
in mnsi , the pinning is observed to be in the @xmath9-directions , which implies @xmath161 , and we will focus on this case from now on . minimizing the free energy with respect to @xmath90 , we find q = , [ eq:2.12 ] which generalizes eq .
( [ eq:2.6b ] ) . choosing @xmath163 and minimizing with respect to @xmath109 we finally have [ eqs:2.13 ] f = -r^2/4u , [ eq:2.13a ] where r = t - t , [ eq:2.13b ] with t = .
[ eq:2.13c ] these results are valid for @xmath19 and @xmath161 . for @xmath164
we expect the pitch vector to move away from @xmath165 towards @xmath166 .
the calculation proceeds as for @xmath19 , except that now the minimization with respect to @xmath123 yields an @xmath11-dependent result . for case
( 2a ) we find [ eqs:2.14 ] _ 3 = 1 & + ( 1 + 2h^2/^2 ) & , [ eq:2.14a ] where ( remember @xmath167 and @xmath168 in the ordered phase for case ( 2a ) ) ^2 = r(b - b_1/2)(t)^2/ua .
[ eq:2.14b ] the helical pitch vector @xmath169 thus moves continuously along the shortest path from its initial value , @xmath170 at @xmath19 , to @xmath166 at @xmath171 , and remains in that position for @xmath172 .
there thus is a second order transition at @xmath171 @xcite that we refer to as the alignment transition .
an inspection shows that case ( 1b ) has a larger free energy for all @xmath173 .
for reference in the next subsection we mention that if one expands the free energy for small values of @xmath123 , and looks for the instability of the solution with @xmath174 , one finds that the latter occurs at @xmath171 , as expected . the circular polarization _
ansatz _ we have used so far explains the two critical fields @xmath14 and @xmath15 observed in mnsi .
however , the solution obtained in this way misses a qualitative feature , as was first pointed out in ref . on symmetry grounds . since in general @xmath176 , the crystal - field contribution to the action , eq .
( [ eq:1.5a ] ) , is not invariant under @xmath177 . as a result
, there is no reason for the @xmath111 and @xmath112-components of @xmath169 to become nonzero at the same value of @xmath11 as @xmath11 is lowered from above , yet the solution constructed in the previous subsection has this property . clearly , this is a result of the fact that our _ ansatz _ with circular polarization possesses cubic symmetry , while the action does not . in general ,
one therefore expects two separate transitions in the vicinity of @xmath14 ; one where the @xmath111-component of @xmath169 becomes nonzero , and a separate one where the @xmath112-component becomes nonzero .
we now show that this expectation is borne out if we allow for a non - circular polarization of the helix , which breaks the cubic symmetry of the _
ansatz_. consider the full eq .
( [ eq:2.11a ] ) , allowing for @xmath178 . minimizing with respect to @xmath124 we see that there are two distinct cases . _ case ( 1 ) : _ : : @xmath141 and @xmath135 arbitrary . minimizing with respect to @xmath135 yields [ eqs:2.15 ] ( ) = .
[ eq:2.15a ] that is , the polarization is in general elliptical .
we have @xmath179 , @xmath180 , which leads to b _ & = & ^2 2 , [ eq:2.15b ] + b_1 & = & ^2 + ^4 . [ eq:2.15c ] considering @xmath181 we find two subcases .
the first one is + _ case ( 1)(i ) : _ ; ; @xmath182 , which implies [ eqs:2.16 ] = ( 0 , , ) , [ eq:2.16a ] and b_1 = ^2- ^4 .
[ eq:2.16b ] the second one is ; ; @xmath183 _ case ( 1)(ii ) : _ ; ; @xmath184 , which implies [ eqs:2.17 ] = ( , 0 , ) , [ eq:2.17a ] and b_1s , a = ^2 ^ 4 .
[ eq:2.17b ] the second case is : : @xmath183 _ case ( 2 ) : _ : : @xmath152 and [ eqs:2.18 ] = 0 .
[ eq:2.18a ] that is , the polarization is circular .
this implies = ( , , ) , [ eq:2.18b ] and b _ & = & 1 - ^4- ^4 , [ eq:2.18c ] + b_1 , & = & 1 - ^2 + ^4[eq:2.18d ] now first consider the case @xmath19
. minimizing with respect to @xmath123 one finds that , for @xmath161 , case ( 2 ) yields the lower free energy.@xcite the physical solution is thus a circularly polarized helix pinned in the @xmath13 directions , and the relaxation of the condition we had imposed in sec .
[ subsubsec : ii.e.1 ] does not change anything . for @xmath160 ,
the physical solution is an elliptically polarized helix pinned in the @xmath185 directions .
next we consider a solution with @xmath186 , which we expect to be stable for sufficiently large @xmath11 .
it is easy to see that case ( 1 ) , which takes advantage of the possibility of an elliptical polarization , has a free energy that is lower than that of case ( 2 ) by a term of @xmath187 everywhere in the conical phase , i.e. , for @xmath98 .
we next look for the instability of the @xmath186 solution at small @xmath11 , which can be found by expanding the action to second order in @xmath123 and looking for the zero of the coefficient of the quadratic term .
as expected , this instability occurs at a field @xmath188 . to leading order in @xmath6 ,
@xmath14 is given by eq .
( [ eq:2.14b ] ) .
the values of @xmath14 for the two cases are different , but the difference is only of @xmath189 and is irrelevant for the following argument .
we now have the following situation . for @xmath173 , the free energy of case ( 1 )
is lower by a term of @xmath187 .
however , we know that at @xmath19 the free energy of case ( 2 ) is lower by a term of @xmath70 .
the two solutions cross at a field @xmath175 given by = 1 - = 1 - o ( ) , [ eq:2.19 ] see fig .
[ fig:6 ] .
-0 mm for case ( 1 ) ( dashed line ) and case ( 2 ) ( dotted line ) , respectively.,title="fig:",width=188 ] at this value of @xmath11 , the state will change discontinuously from an elliptically polarized helix with a pitch vector given by either eq.([eq:2.16a ] ) or ( [ eq:2.17a ] ) to a circularly polarized one with a pitch vector given by eq .
( [ eq:2.18b ] ) .
we now have the following progression of phases and phase transitions as the magnetic field is lowered from a value greater than @xmath15 : @xmath101 : field - polarized state , no helix .
@xmath190 : second - order transition to a conical state with an elliptically polarized helix , @xmath191 .
@xmath192 : second - order transition to a conical state with elliptical polarization as above , but @xmath193 .
@xmath123 increases from zero with decreasing @xmath11 .
@xmath194 $ ] : first - order transition to a conical state with circular polarization and @xmath195 .
@xmath123 increases from its value at @xmath175 with decreasing @xmath11 .
@xmath196 : system reaches helical state with circular polarization and @xmath197 . the phase diagram is thus predicted to have the structure shown in fig.[fig:7 ] , with the second order alignment transition at @xmath14 followed by a first - order polarization ( and re - alignment ) transition at @xmath175 .
the latter has so far not been observed experimentally .
-0 mm -@xmath18 plane showing the predicted polarization transition at @xmath175 ( first order , dashed line ) in addition to the second - order transitions at @xmath14 and @xmath15 .
the nature of the transition from the conical phase to the a - phase , or the a - phase to the paramagnetic phase , has not been investigated so far . the dotted line denotes the crossover from a field - polarized paramagnet to a field - polarized ferromagnet.,title="fig:",width=226 ] it is an explicit realization of the type of transition first predicted by walker@xcite on symmetry grounds .
it needs to be stressed that the detailed features of this transition are restricted by our _ ansatz _ , eqs.([eqs:2.1 ] ) ; the states obtained are not true minima of the action . however , the basic physical idea , which is expected to be realized in the true ground state as well , is as follows . in the conical phase , where the helical pitch vector is aligned with the magnetic field , the system can take advantage of the lack of cubic symmetry of the action , eq .
( [ eq:1.5a ] ) , by forming a helix with a non - circular polarization .
this leads to an energy gain , compared to a circularly polarized state , of order @xmath198 .
however , it forces the pitch vector into either the @xmath199 or @xmath200 plane , i.e. , case ( 1 ) above . at low fields ,
these states have a free energy that is larger by a term of @xmath70 than the states with the pitch vector along a cubic diagonal , which requires a circular polarization .
the competition between these two effects leads to the first order transition at @xmath14 , where both the polarization and the orientation of the pitch vector change discontinuously .
experimentally , the a - crystal is found to be pinned very weakly , and theoretical considerations conclude that the pinning potential is only of @xmath187.@xcite this is a consequence of the six - fold rotation symmetry of the a - crystal .
similarly one expects the ( average ) direction of the fluctuating defect lines in the nfl region to be pinned only very weakly .
we have treated the phase transitions that we have discussed in this section within classical mean - field theory , and the question arises what changes will result from the consideration of fluctuations , classical or quantum .
one example of the effects of fluctuations is the tricritical point and the associated tricritical wings shown in fig .
[ fig:3 ] , which result from quantum fluctuations that can be taken into account within a renormalized mean - field theory.@xcite elsewhere in the phase diagram , fluctuations are also of qualitative importance .
consider , for instance , the transition from the conical phase to the field - polarized phase at the critical field @xmath15 , see fig .
[ fig:1 ] .
the transition is characterized by the vanishing of the one - dimensional order parameter @xmath109 , the amplitude of the helix , see eqs.([eqs:2.6 ] ) .
the action is invariant under @xmath201 , and we therefore expect this transition at nonzero temperature to be in the universality class of the classical ising model.@xcite at zero temperature the statics and the dynamics will couple,@xcite and one expects the quantum phase transition to be described by hertz s model@xcite with a scalar order parameter .
goldstone s theorem states that if a continuous symmetry described by a lie group @xmath202 is spontaneously broken , with the remaining subgroup in the broken - symmetry phase being @xmath11 ( not to be confused with the magnetic field ) , then in the ordered phase there are @xmath203 soft or massless modes , with @xmath203 equal to the dimensionality of the coset space @xmath204.@xcite all of the ordered states discussed in sec . [ sec : ii ] break a continuous symmetry , and therefore there must be one or more soft modes in the ordered phase.@xcite the number and functional form of the goldstone modes in the various phases can be determined from general arguments . in this section
we will do so , and we will augment these arguments by explicit calculations in some cases where this is feasible . via a coupling to the conduction electrons , the goldstone modes have interesting consequences for various observables ; this is the topic of sec .
[ sec : iv ] .
to zeroth order in @xmath6 the system is a ferromagnet , see sec.[subsec : ii.a ] .
the relevant symmetry is the rotation group , @xmath205 , which in the ordered state is spontaneously broken to @xmath206 .
hence there are @xmath207 goldstone modes , the well - known ferromagnons ( fm ) .
their dispersion relation for small wave numbers is _
( k ) = dk^2 , [ eq:3.1 ] with @xmath208 the spin wave stiffness , which depends on the magnetization and vanishes as the magnetization goes to zero .
the easiest way to derive this result is to consider a nonlinear sigma - model - version of the heisenberg action.@xcite neglecting fluctuations of the amplitude of the magnetization , which can be shown to be massive , one parameterizes the order parameter ( x ) = m_0 ( c _
1(x ) + _ 2(x ) + ) [ eq:3.2 ] and expands the action to bilinear order in @xmath209 . if we neglect the small relativistic term with coupling constant @xmath50 in eq .
( [ eq:1.3 ] ) , the resulting quadratic form has two identical eigenvalues = ( ak^2 + h ) [ eq:3.3 ] for @xmath19 , one has @xmath210 , which reflects the two goldstone modes .
physically , the homogeneous transverse magnetic susceptibility diverges .
this is the static manifestation of the spontaneously broken continuous symmetry .
determining the dynamics requires an additional step .
one either needs to solve an appropriate langevin equation within a classical context,@xcite or treat the problem quantum mechanically.@xcite either way one obtains eq.([eq:3.1 ] ) with @xmath211 .
when the spin - orbit coupling is taken into account , we have the various phases involving helical spin textures discussed in sec .
[ subsec : ii.a ] . for all of these phases
the relevant symmetry is the translational one .
let @xmath18 be the lie group of one - dimensional translations .
then the action is invariant under @xmath212 .
consider the aligned conical state ( acs ) discussed in sec .
[ subsec : ii.b ] , from which the unpinned helical phase emerges as a special case at @xmath19 .
@xmath213 is broken down to @xmath214 ( translations in the two directions perpendicular to @xmath215 ) , so we expect one goldstone mode . at @xmath19 ,
one expects the soft fluctuations in the ordered phase to be phase fluctuations of the helix , and one might naively expect them to be governed by an action @xmath216 , with @xmath217 the phase variable , which would lead to a soft eigenvalue proportional to @xmath218 . however , the ordered state is also invariant under rotations of @xmath0 , which can be written as a phase fluctuation with a nonvanishing gradient , so this can not be the correct answer .
the lowest - order term allowed by rotational symmetry that involves the gradients perpendicular to @xmath0 is of the form @xmath219 , with @xmath51 a generalized phase variable , and this leads to an eigenvalue , or inverse order parameter susceptibility , proportional to the term under the square root in eq.([eq:1.1]).@xcite the dynamics again require additional considerations , which lead to a resonance frequency that is proportional to the square root of the inverse susceptibility , unlike the ferromagnetic case , and this is expressed in eq.([eq:1.1]).@xcite these results were first obtained by means of an explicit calculation for both classical and quantum helimagnets in ref . . a magnetic field breaks the rotational invariance ,
so a @xmath220-term will be present in the dispersion relation .
the prefactor is expected to be analytic in @xmath11 , and we thus expect for the dispersion relation of this `` conimagnon '' , the goldstone model of the acs , _
( k ) = , [ eq:3.4 ] with @xmath221 . for the action to @xmath65 , eqs .
( [ eqs:2.6 ] , [ eqs:2.7 ] ) constitute an exact saddle - point solution , and we can perform an explicit calculation of the goldstone mode .
a complete parameterization of fluctuations about the saddle point can be written ( x ) & = & ( m_0 + m_0(x ) ) ( c _
1(x ) + _ 2(x ) + ) + & & + ( c ( qz + _
0(x ) ) + ( qz + _
0(x ) ) + ( x ) ) .
+ [ eq:3.5 ] the first term is the nonlinear sigma model for the homogeneous magnetization from sec .
[ subsec : iii.a ] , and the second one parameterizes fluctuations of the helix in terms of an amplitude modulation , a phase @xmath222 , and a third component @xmath223 .
the amplitude fluctuations one expects to be massive , and an explicit calculation confirms this , so we drop @xmath224 and @xmath225 .
the field @xmath226 is conveniently written ( x ) = _ + ( x)(q ) + _ -(x)(q ) , [ eq:3.6 ] which ensure that @xmath227 and @xmath222 at zero wave number both correspond to @xmath4 at wave number @xmath90 .
double counting is avoided by restricting the theory to wave numbers small compared to @xmath90 .
if we use eq.([eq:3.5 ] ) in the action to @xmath65 , eqs.([eq:1.3],[eq:1.4a ] ) , and expand to bilinear order in the fluctuations @xmath228 , we obtain a gaussian action ^(2 ) = _ k _ i=1 ^ 5 _ i(k)_ij(k)_j(-k ) .
[ eq:3.7 ] a sketch of the derivation , and the explicit form of the matrix @xmath229 , are given in appendix [ app : b ] .
of the five eigenvalues , one goes to zero for @xmath230 , in agreement with the expectation from the symmetry arguments given above .
it takes the form [ eqs:3.8 ] _ 1 = k_z^2 + _ ^2 + _ ^4 , [ eq:3.8a ] with @xmath231 , and coefficients & = & m_1 ^ 2 , [ eq:3.8b ] + & = & , [ eq:3.8c ] + & = & m_1 ^ 2 .
+ [ eq:3.8d ] here we have defined @xmath232 .
this result is consistent with eq .
( [ eq:3.4 ] ) . for @xmath19 , which implies @xmath233
, it reduces to the helimagnon result of ref .. in addition , there are four massive eigenvalues that appear in pairs . at zero wave number , they are [ eqs:3.9 ] _ 2 = _ 3
_ = m_1 ^ 2 [ 1 + m_0 ^ 2 + o(m_0 ^ 4 ) ] , [ eq:3.9a ] + _ 3 = _ 4 _
= m_0 ^ 2 [ 1 + m_1 ^ 2 ] + o(m_0 ^ 4 ) .
[ eq:3.9b ] we recognize @xmath234 as representing the massive helimagnon modes,@xcite modified by the the presence of @xmath85 , and @xmath235 as the massive ( due to the presence of a magnetic field ) ferromagnons , eq .
( [ eq:3.3 ] ) , modified by the presence of @xmath109 .
the helical states that have been proposed as candidates for the a - phase and were discussed in sec .
[ subsec : ii.e ] are not saddle - point solutions of the action , which precludes a model calculation of the goldstone modes resulting from this type of order . however , assuming that the order is stabilized by some mechanism , the functional form of the soft modes can still be determined by symmetry arguments analogous to those put forward in the previous subsection .
consider a skyrmion lattice state .
the state described by eq .
( [ eq:2.9 ] ) is invariant only under translations in one direction , viz . , the direction perpendicular to the plane of helices .
the same is true for any state that is characterized by columnar order , so this property does not depend on the precise nature of the skyrmions .
any such state will thus have @xmath236 goldstone modes .
this was to be expected : since the skyrmions form a two - dimensional lattice , there should be two generalized phonon modes , namely , a compression mode and a shear mode . in zero magnetic field , the energy would still be invariant under global rotations of the skyrmion lattice .
hence , the soft eigenvalue can have no @xmath237 contribution . for @xmath81
this is no longer true , and we thus expect = k_^2 + k_z^4 + k_z^2 , [ eq:3.10 ] for the soft eigenvalue , with @xmath238 , and _
( k ) = , [ eq:3.11 ] for the dispersion relation , with @xmath239 and @xmath240 . if the nfl region can be interpreted as a molten a - crystal , see sec.[subsec : ii.d ] then one of the two goldstone modes , the compression mode , will persist as long as their is columnar structure .
this is analogous to the fact that longitudinal phonons , or ordinary sound , exist in a liquid , whereas in a crystal one has transverse phonons or shear modes in addition .
the nfl region extends to @xmath19 , where the dispersion relation of the compressional goldstone mode is given by eq .
( [ eq:3.11 ] ) with @xmath241 to first order in @xmath6 .
now consider the crystal - field terms in the action that first appear at @xmath70 , eq .
( [ eq:1.5a ] ) . for simplicity ,
let us consider the term with coupling constant @xmath242 .
it breaks rotational invariance , which invalidates the argument that leads to the absence of a @xmath36-term in the soft mode resonance frequency .
the system must remain stable regardless of the sign of @xmath242 , and we thus expect for the dispersion relation of the helimagnons in the pinned helical state _
( k ) = , [ eq:3.12 ] with @xmath243 , which replaces eq . ( [ eq:1.1 ] ) . for the more general model given by eqs .
( [ eq:1.2 ] ) - ( [ eqs:1.5 ] ) , @xmath244 , @xmath245 , and @xmath11 will also contribute to the elastic constant @xmath246 .
we now check this expectation by means of an explicit calculation . for the model with only the first of the crystal - field terms present
, we have an exact saddle - point solution , namely , [ eqs:3.13 ] _ ( x ) = m_1 , [ eq:3.13a ] with @xmath126 , @xmath247 , and @xmath0 a _ dreibein _ , m_1 = , [ eq:3.13b ] where r = t - cq/2 , [ eq:3.13c ] and & = & q ( 1,1,1)/ & + ( 1,0,0 ) & , [ eq:3.13d ] + q & = & c/2(a+b/3 ) & + c/2a & . [ eq:3.13e ]
the parameterization of fluctuations about this state is given by the second term in eq .
( [ eq:3.5 ] : ( x ) & = & ( m_1 + m_1 ) , [ eq:3.14 ] with @xmath226 given by eq . ( [ eq:3.6 ] ) .
we again drop the massive amplitude fluctuations and expand the action to quadratic order in the phase fluctuations .
the gaussian action is of the form ^(2)[_i ] = _ k _ = 0 , _
( k ) _ ( k ) _ ( k ) .
[ eq:3.15 ] the explicit form of the matrix @xmath248 depends on the sign of @xmath242 ; it is given explicitly in appendix [ app : c ] .
an inspection of the eigenvalues shows that in either case there is one eigenvalue @xmath249 that vanishes as @xmath230 and hence represents the goldstone mode , as expected from the symmetry argument given above . to order @xmath250 and @xmath251
we find = ak_||^2 + bk_^2 + + ( a+b)k_^4/q^2 & -50pt , + ( a + b/3)k_||^2 + b k_^2 + ( a+b+o(^4 ) ) k_^4/q^2 + & -50pt .
[ eq:3.16 ] in the case @xmath252 we have neglected terms proportional to @xmath253 in the prefactor of @xmath254 . this result agrees with the functional form obtained by symmetry arguments alone , eq .
( [ eq:3.1 ] ) , and for @xmath255 it correctly reduces to the result for the isotropic model , ref . and
( [ eq:1.1 ] ) .
we can now summarize the results for the single - helix phases discussed above as follows . in the helical and conical phases , including pinning effects , there is a single goldstone mode with a resonance frequency _ 0(k ) = , [ eq:3.17 ] with @xmath256 small compared to @xmath34 and @xmath35 .
this comprises eqs .
( [ eq:3.4 ] ) and ( [ eq:3.12 ] ) .
in the a - phase , pinning effects are weaker than in the helical phase due to the hexagonal nature of the skyrmion lattice , see sec .
[ subsubsec : ii.e.4 ] above .
when this weak pinning is taken in to account , the goldstone mode is thus given by eq .
( [ eq:3.11 ] ) with @xmath257 .
while in mnsi the a - phase is observed only in an external magnetic field , there is no intrinsic reason why in some other system it could not be stable in a zero field .
the goldstone modes in such a system would be given by eq.([eq:3.11 ] ) with an extremely small @xmath258 . by the same argument ,
in the nfl region at @xmath19 we expect a goldstone mode given by eq .
( [ eq:3.11 ] ) with @xmath259 . to summarize
, we have found that the ( single ) goldstone mode in the pinned helical and conical phases is given by eq .
( [ eq:3.17 ] ) . in the pinned helical phase the parallel direction
is determined by the crystal - field effects that pin the helix ; in the conical phase , it is the direction of the magnetic field ( which we have chosen to be the @xmath120-direction for all of our considerations ) .
the elastic constant @xmath260 is small compared to the other elastic constants . in the pinned helical phase it is of @xmath65 , and in the conical phase it is of @xmath261 .
by contrast , @xmath34 and @xmath35 are of @xmath62 and @xmath79 , and of @xmath262 , respectively . in the a - phase
, there are two goldstone modes whose dispersion relation is given by eq .
( [ eq:3.11 ] ) .
the elastic constant @xmath258 is small of @xmath261 and @xmath187 compared to the others . in the nfl region , the single goldstone mode is also given by eq .
( [ eq:3.11 ] ) , with @xmath259 . finally , in the perpendicular conical state discussed in appendix [ app : a ] there is a single goldstone mode that in the absence of crystal - field effects is given by eq .
( [ eq : a.3 ] ) .
if pinning by crystal - field terms is taken into account there is an additional term @xmath263 under the square root , with @xmath264 .
all of these results are also summarized in table [ tab:1 ] in sec .
[ sec : v ] . the structures of the various goldstone modes , and the fact that the various elastic coefficients have very different magnitudes , leads to the formation of different temperature regimes that are dominated by different physics .
we now explain this using the generalized helimagnons , eq .
( [ eq:3.17 ] ) , as an example ; the argument for the other cases works analogously .
as far as the coupling of the magnetic goldstone mode to the electronic degrees of freedom is concerned , the resonance frequency @xmath265 scales as the temperature , @xmath266 . if we scale @xmath267 with @xmath268 and @xmath269 with @xmath270 , we obtain _ 0(k ) = t , [ eq:3.18 ] where @xmath271 now denotes the scaled , dimensionless , wave number .
for @xmath272 the symmetry - breaking @xmath36 term is negligible , and the goldstone mode is effectively what it would be in a rotationally invariant system . in this regime
the physics is dominated by universal hydrodynamic effects that are independent of the microscopic details of the solid and analogous to the corresponding effects in liquid crystals . in the opposite limit , @xmath273 , the crystal - field effects due to the underlying ionic lattice , or the external magnetic field , if present , dominate and the goldstone mode has the same functional form as ( anisotropic ) acoustic phonons . due to the smallness of @xmath260
the universal hydrodynamic regime is sizable , and it is in this region that the most interesting effects of the magnetic order manifest themselves in the electronic properties of the system .
this is true especially if the goldstone mode appears in zero magnetic field and the pinning is very small , such as in the nfl region or a ( so far hypothetical ) a - phase in zero field .
via a coupling to the conduction electrons , the goldstone modes derived in the preceding section influence the electronic properties of the helical magnet . in this section
we derive the consequences for the specific heat , the single - particle relaxation rate , and the thermal and electrical resistivities . in all cases
we consider the contribution of the goldstone mode in isolation ; it comes in addition to all other contributions to these observables .
any well - defined ( i.e. , not overdamped ) excitation with a dispersion relation @xmath274 yields a contribution to the specific heat @xmath275 given by c(t ) = _ k ( k ) n_((k ) ) .
[ eq:4.1 ] here @xmath276 is the bose distribution function , @xmath277 is the system volume , and we use units such that @xmath278 .
this allows one to determine the contributions to the specific heat by the various goldstone modes .
we first consider the helical and aligned conical phase .
the dispersion relation is given by eq .
( [ eq:3.17 ] ) . performing the integral in eq.([eq:4.1 ] )
yields c(t ) = t^3/c _ & + t^2/ & , [ eq:4.2 ] the universal hydrodynamic result , @xmath279 , was first derived in ref . .
it is subleading to , but distinct from , the fermi - liquid result @xmath280 . at
asymptotically low temperatures it crosses over to a @xmath213 behavior consistent with the acoustic - phonon - like dispersion relation in either the pinned helical phase or the aligned conical phase at asymptotically small wave numbers . in the skyrmion - lattice state
we find from eq .
( [ eq:3.11 ] ) in conjunction with eq .
( [ eq:4.1 ] ) c(t ) = t^3/c_c_z & + t^5/2/c_c_z^1/4 & .
[ eq:4.3 ] the temperature dependence of the single - particle relaxation rate , as well as that of the electrical transport relaxation rate or the electrical resistivity , can be obtained by using the results of refs .. the former paper obtained the following expression for the single - particle relaxation rate , & = & 2 _ -^ _ p v(p - k;k , p;u ) + & & ( u - _ 1(p ) ) . [ eq:4.4 ] here the quasiparticle momentum @xmath281 is taken to be on the 1-fermi surface , @xmath282 , with [ eqs:4.5 ] _
1,2(k ) = , [ eq:4.5a ] which is separated from the 2-fermi surface , @xmath283 , by twice the stoner gap @xmath249 . here _ k = _ k - , [ eq:4.5b ] with _
k = k^2/2 + ( k_x^2 k_y^2 + k_y^2 k_z^2 + k_z^2 k_x^2 ) [ eq:4.5c ] an electronic energy - momentum relation consistent with a cubic crystal .
the dimensionless parameter @xmath284 is a measure of deviations from a nearly - free electron model .
@xmath285 is the spectrum of an effective potential given by [ eqs:4.6 ] v(k;p_1,p_2 ) = ^2(k)(k , p_1)(-k , p_2 ) , [ eq:4.6a ] with vertices ( k , p ) & = & + & & + o(k^2 ) , [ eq:4.6b ] and @xmath286 the soft - mode susceptibility , ( k ) = .
[ eq:4.6c ] here @xmath287 is the resonance frequency , which is equal to the frequency given in eq .
( [ eq:3.4 ] ) , ( [ eq:3.11 ] ) , ( [ eq:3.12 ] ) , or ( [ eq : a.3 ] ) , depending on the phase under consideration . the electrical transport relaxation time , which determines the electrical resistivity ,
is effectively given by averaging a similar expression over the fermi surface,@xcite & = & _ -^ _ p , k + & & v(p - k;k , p;u)(u - _ 1(p))(_1(k ) ) .
+ [ eq:4.7 ] note the additional , compared to eq .
( [ eq:4.4 ] ) , factor of @xmath288 under the integral in eq .
( [ eq:4.7 ] ) .
this is characteristic of the description of electrical transport in a boltzmann approximation and leads to a temperature dependence of the electrical resistivity that is different from that of the single - particle relaxation rate.@xcite in contrast , in a boltzmann description of thermal transport this additional factor is absent , and the temperature dependence of the thermal conductivity is given by that of the single - particle relaxation rate.@xcite we note that these expressions for @xmath289 and @xmath290 vanish as @xmath291 . while they give the leading asymptotic temperature dependence for the relaxation rates in helimagnets , they therefore can not be used to obtain the corresponding results in the ferromagnetic limit .
the asymptotic low-@xmath18 behavior of the rates in ferromagnets is qualitatively different and briefly treated in appendix [ app : d ] . in the context of helimagnets , for a non - spherical fermi surface ( i.e. , @xmath292 in eq .
( [ eq:4.5c ] ) ) , and for generic wave vectors @xmath281 , the temperature scaling behavior of the single - particle relaxation rate @xmath293 , the thermal resistivity @xmath38 , and the electrical resistivity @xmath25 can be represented schematically by the expressions [ eqs:4.8 ] & ~ & _ ~dp_||dp_^2 + & & .
[ eq:4.8a ] + _ & ~ & dp_||dp_^2 + & & .
[ eq:4.8b ] the resonance frequency always scales as the temperature , @xmath266 , and the temperature dependence of the relaxation rates thus is determined by how the momentum components scale with temperature . for the helical and conical phases we have , from eq .
( [ eq:3.17 ] ) , @xmath294 , and @xmath295 and @xmath296 at asymptotically low and intermediate temperatures , respectively .
this yields the following temperature dependence for the single - particle relaxation rate and the thermal resistivity : _
t^3 & + t^3/2 & .
[ eq:4.9 ] the corresponding result for the electrical resistivity is _
t^5 & + t^5/2 & .
[ eq:4.10 ] in a vanishing external field , and in a temperature regime where pinning effects are not relevant , we recover the @xmath297 and @xmath298 behavior for @xmath37 and @xmath25 , respectively , of ref . and .
for a skyrmion lattice , we obtain by using eq .
( [ eq:3.11 ] ) _
t^3 & + t^2 & .
[ eq:4.11 ] for the single - particle rate , and _ t^5 & + t^3 & .
[ eq:4.12 ] for the electrical resistivity .
the preceding results hold for clean systems . in the presence of quenched disorder , elastic scattering of the conduction electrons
leads to profound effects that manifest themselves in the transport properties .
one needs to distinguish between the strong - disorder regime , where the transport is diffusive , and the weak - disorder regime , where it is ballistic . in a fermi liquid , these two regimes are characterized by @xmath299 and @xmath300 , respectively.@xcite in a helical magnet , the weak - disorder or ballistic regime is characterized by@xcite 1 , [ eq:4.13 ] where @xmath301 is the elastic relaxation time . in this regime
there is an additional contribution @xmath302 to the relaxation rates that is qualitatively the same for both the single - particle rate and the electron transport rate,@xcite and thus provides the temperature dependence of the corrections to both the electrical and thermal resistivities .
it was shown in refs .
that , for temperature scaling purposes , this contribution can be represented by ( 1/ ) & & _ _ + & & -40pt ~du n_(u / t ) dp_||
dp_^2 ( u - _ 0(p ) ) . + [ eq:4.14 ] from eq .
( [ eq:4.14 ] ) we see that at temperatures where pinning effects are not important , generalized helimagnons lead to @xmath303 , a result first obtained in ref . .
at asymptotically low temperatures , characterized by @xmath304 , one finds a @xmath214-behavior . in the pinned helical phase the crossover temperature between these two regimes
is determined by the strength of the crystal - field effects ; in the conical phase the magnetic field also cuts off the universal hydrodynamics @xmath18-behavior . for the a - phase , eq .
( [ eq:3.11 ] ) yields ( 1/ ) t^2 & + t^3/2 & .
[ eq:4.15 ] the pinning effects in the a - phase are very weak due to the hexagonal nature of the skyrmion lattice , with @xmath258 only of @xmath187 , see secs.[subsubsec : ii.e.4 ] and [ subsubsec : iii.d.4 ] .
the size of the asymptotic region is therefore likely to be dominated by the @xmath11-dependence of @xmath258 .
whether or not the universal hydrodynamic @xmath297-behavior is observable in the a - phase ( there currently are no experimental indications that it is ) would require a detailed quantitative analysis that goes beyond the scope of the current paper .
the preceding result is also of interest with respect to the non - fermi - liquid region shown in fig .
[ fig:2 ] , which is _ not _ a phase with long - range order , but where the electrical conductivity shows a pronounced @xmath297-behavior.@xcite an explanation that has recently been proposed@xcite is as follows .
the @xmath297-behavior derived above is a consequence of the structure of the goldstone modes due to columnar fluctuations , eq .
( [ eq:3.11 ] ) , in conjunction with weak quenched disorder . in the a - phase , which displays long - range order in the form of a skyrmion lattice , there are two such goldstone modes , see sec .
[ subsec : iii.c ] .
if the nfl region can be interpreted as a melted skyrmion lattice , then the resulting skyrmion fluid will still have one goldstone mode with the same structure , namely , the compression mode mentioned in sec .
[ subsec : iii.c ] . weak quenched disorder will then still produce a contribution to the electrical resistivity , as well as to the single - particle relaxation rate , that is proportional to @xmath297 in a pre - asymptotic region .
the nfl region is observed to extend to a vanishing external magnetic field , so the low-@xmath18 boundary of the universal hydrodynamic region is determined by the pinning effects , which are very weak , see secs.[subsubsec : ii.e.4 ] and [ subsubsec : iii.d.4 ] . the universal hydrodynamic @xmath297 behavior is therefore expected to extend to very low temperatures .
a remaining question is the size of the prefactor , which in a bare theory is expected to be small due to the long length scale set by the skyrmion lattice .
the resolution proposed in ref .
is that mode - mode coupling effects drastically enhance the magnitude of the effect , in analogy to what is believed to happen in the blue phases of liquid crystals.@xcite
c c|c|c||c|c|c|c & & & & pinned helical & a - phase & pcs & nfl + & & & & /conical & @xmath305 & @xmath306 & + & & & & & & & + & & goldstone & @xmath287 & @xmath307 & @xmath308 & @xmath309 & @xmath310 + & & & & & & & -374pt + & & modes & # & 1 & 2 & 1 & 1 + & & & & & & & + & & & & & & & + & & & @xmath311 & @xmath214 & @xmath298 & @xmath214 & + & & @xmath312 & & & & -276pt & @xmath298 + & & & @xmath313 & @xmath213 & @xmath213 & @xmath298 & + & & & & & & & -422pt + & & @xmath314 , & @xmath311 & @xmath297 & @xmath214 & @xmath297 & + -10pt clean & & & & & & -276pt & @xmath214 + & & @xmath315 & @xmath313 & @xmath213 & @xmath213 & @xmath214 & + & & & & & & & -422pt + & & & @xmath311 & @xmath298 & @xmath213 & @xmath298 & + & & @xmath316 & & & & -276pt & @xmath213 + & & & @xmath313 & @xmath317 & @xmath317 & @xmath213 & + & & & & & & + & & & & & & & + & & @xmath318 , & @xmath319 & @xmath18 & @xmath297 & @xmath18 & + -20pt weak & -20pt disorder & & & & & -276pt & @xmath297 + & & & & & & & + in summary , we have given a comprehensive description of all phases in dzyaloshinsky - moriya helical magnets where long - range order has been observed .
these include , the pinned helical phase at weak magnetic fields , the conical phase at higher magnetic fields , and the a - phase at intermediate magnetic fields and temperatures close to the critical temperature .
we have shown that the system goes from the conical phase at high magnetic field to the pinned helical phase at low magnetic field via two distinct phase transitions ; a second - order transition where the orientation of the helix changes smoothly , followed by a first - order transition where both the orientation and the polarization of the helix change discontinuously . for the a - phase we have considered the recent interpretation , based on the observation of a six - fold symmetry in the neutron scattering signature , as a hexagonal lattice of skyrmionic line defects .
we have also discussed the perpendicular conical state , which had been discussed earlier as a possible realization of the a - phase .
in addition , we have discussed a proposal for the interpretation of the non - fermi - liquid ( nfl ) region in the disordered phase , which does not display long - range magnetic order but has many features in common with the a - phase .
for all of these states we have determined the number and nature of the goldstone modes , and the temperature dependencies of various observables that result from the scattering of the conduction electrons by these excitations .
if the state in question can be described by an exact saddle - point solution of a model hamiltonian we have provided an explicit calculation ; in other cases we have used symmetry arguments to determine the functional form of the goldstone modes .
the results are summarized in table [ tab:1 ] .
the goldstone modes show unusual anisotropic frequency - momentum relations that result from helical spin structures in the various states and are reminiscent of the goldstone modes in smectic and cholesteric liquid crystals .
the specific heat @xmath312 , the single - particle relaxation rate @xmath314 and the thermal resistivity @xmath315 , and the electrical resistivity @xmath316 , all display temperature dependencies that are distinct from , and in some cases stronger than , those in a fermi liquid .
the most remarkable result is a non - fermi liquid @xmath297 behavior of the electrical resistivity in certain temperature regimes in the a - phase and in the nfl region , which is proposed as an explanation of the observed enigmatic properties of the dm magnet mnsi .
this work was initiated at the aspen center for physics , and supported by the national science foundation under grant nos .
dmr-09 - 29966 and dmr-09 - 01907 .
guided by earlier experimental results that did not reveal the six - fold symmetry discovered by mhlbauer et al.,@xcite grigoriev et al.@xcite had proposed a perpendicular conical state ( pcs ) to represent the a - phase , i.e. , a spin configuration of the form ( x ) = m_0 + m_1 [ eq : a.1 ] with @xmath0 perpendicular to @xmath42 and @xmath126 , @xmath320 , and @xmath169 forming a _
dreibein_. this state is not a saddle point either , and ref
. found it energetically unfavorable compared to the conical and skyrmion lattice states .
however , it can be stabilized , at least in principle , and we therefore include it in our discussion . ref . proposed that the pcs is stabilized due to a gap in the helimagnon excitation spectrum , which in turn had been proposed theoretically in ref . .
such a gap is at odds with goldstone s theorem as well as with the calculation in ref . .
a different possibility , which is consistent with symmetry considerations , is a large negative value of the parameter @xmath52 in eq .
( [ eq:1.3 ] ) , which stabilizes the pcs in a parameter regime close to where the a - phase is observed , see fig .
[ fig:8 ] .
while such a large value of @xmath52 is not realistic , this illustrates that the pcs can be stabilized by terms in the action that are allowed by symmetry , and it therefore is useful to determine the properties of this state .
let us now consider the goldstone modes in the pcs , using the symmetry arguments explained in sec .
[ subsec : iii.a ] . for definiteness
, we assume that the @xmath0-vector points in the @xmath111-direction .
the ordered state is still invariant under translations in the @xmath112 and @xmath120-directions , so we have @xmath321 goldstone mode . with the magnetic field @xmath11 in the @xmath120-direction , the state is still invariant under rotations of @xmath0 in the @xmath111-@xmath112-plane .
the eigenvalue therefore can not have a @xmath322 term .
the soft eigenvalue therefore must have the form = k_x^2 + ( k_y^4 + k_z^4 ) + k_z^2 , [ eq : a.2 ] where @xmath238 for small @xmath11 .
the structure of the dynamics will be the same as in the aligned conical case , and we thus have a dispersion relation _
( k ) = , [ eq : a.3 ] with @xmath323 and @xmath324 and @xmath35 constant for @xmath82 . here
we neglect pinning effects , which lead to a @xmath322-term under the square root at asymptotically low wave numbers .
-0 mm . shown are the regions of stability for various phases for the action shown in eq .
( [ eq:1.3 ] ) with @xmath325 , @xmath326 , @xmath327 , and @xmath328.,title="fig:",width=294 ] for the specific heat contribution we find , from eqs .
( [ eq:4.1 ] ) and ( [ eq : a.3 ] ) , c(t ) = t^5/2/(c_xc_z)^1/2c_^1/4 & + t^2/ & .
[ eq : a.4 ] this result is different from the one for the skyrmionic goldstone mode in sec .
[ subsec : iii.c ] .
measurements of the specific heat therefore offer a way to distinguish between these two states short of a direct determination of the spin structure .
using eq .
( [ eq : a.3 ] ) in an obvious generalization of eqs .
( [ eqs:4.8 ] ) we find for the single - particle relaxation rate and the thermal resistivity _
t^2 & + t^3/2 & , [ eq : a.5 ] and for the electrical resistivity _
t^3 & + t^5/2 & .
[ eq : a.6 ]
here we derive the gaussian fluctuation action in the conical phase , eq.([eq:3.7 ] ) . the equation of state for this phase , eqs .
( [ eqs:2.6 ] , [ eqs:2.7 ] ) , can be written t - aq^2 + u(m_0 ^ 2 + m_1 ^ 2 ) = 0 , [ eq : b.1 ] with @xmath90 and @xmath85 given by eqs . ( [ eq:2.6b ] ) and ( [ eq:2.6c ] ) , respectively . inserting eq .
( [ eq:3.10 ] ) into the action , eqs.([eq:1.3 ] , [ eq:1.4a ] ) , dropping the amplitude fluctuations @xmath224 and @xmath225 , and expanding to bilinear order in the phases yields eq.([eq:3.7 ] ) with \(k ) = ( ccccc m_1 ^ 2k^2 & -i m_1
^ 2 k_y & i m_1 ^ 2 k_x & 0 & 0 + i m_1 ^ 2 k_y & m_1 ^ 2 ( 1 + m_0 ^ 2 + k^2 ) & -i m_1 ^ 2 k_z & 0 & m_0 ^ 2 m_1 ^ 2 + -i m_1 ^ 2 k_x & i m_1 ^ 2 k_z & m_1 ^ 2 ( 1 + m_0 ^ 2 + k^2 ) & m_0 ^ 2 m_1 ^ 2 & 0 + 0 & 0 & m_0 ^ 2 m_1 ^ 2 & m_0 ^ 2 ( 1 + m_1 ^ 2 + k^2 ) & 2i m_0 ^ 2 k_z + 0 & m_0 ^ 2 m_1 ^ 2 & 0 & -2i m_0 ^ 2 k_z & m_0 ^ 2 ( 1 + m_1 ^ 2 + k^2 ) ) .
[ eq : b.2 ] a determination of the eigenvalues yields eqs .
( [ eqs:3.8 ] , [ eqs:3.9 ] ) .
here we derive the gaussian fluctuation action in the pinned helical phase , eq .
( [ eq:3.15 ] ) . dropping the amplitude fluctuations , and expanding to linear order in the phase fluctuations , eq .
( [ eq:3.14 ] ) becomes [ eqs : c.1 ] ( x ) = m_(x ) + ( x ) , [ eq : c.1a ] with ( x ) = m_1 .
[ eq : c.1b ] @xmath329 is an exact saddle point of the action , eqs.([eq:1.2 ] ) - ( [ eqs:1.5 ] ) with @xmath330 , so the terms linear in @xmath331 vanish .
we now consider terms bilinear in @xmath332 and first concentrate on the gradient - free terms .
neglecting rapidly fluctuating fourier components proportional to @xmath333 with @xmath334 one finds ^(2)[]/m_1 ^ 2 & = & dx_0(x)_0(x ) + & & + dx _ = _ ( x)_(x ) + & & -bq^2 dx _ = _
i=1 ^ 3 e_^i q_i^3_0(x)_(x ) + [ eq : c.2 ] an explicit calculation shows that [ eqs : c.3 ] _
i=1 ^ 3 q_i^2 & = & 1 - f ( ) , [ eq : c.3a ] + _ i=1 ^ 3 q_i^4 = f ( ) , [ eq : c.3b ] with f ( ) = _ 1 ^ 4 + _ 2 ^ 4 + _ 3 ^ 4 , [ eq : c.3c ] where @xmath335 are the direction cosines of @xmath169 .
furthermore , _
i=1 ^ 3 e_^i ( q_i)^3 = 0 [ eq : c.3d ] for @xmath336 and for both @xmath157 and @xmath151 .
using the equation of state , eqs .
( [ eq:3.13b ] , [ eq:3.13c ] ) , we see that the @xmath222-@xmath222 vertex and the @xmath222-@xmath337 vertices all vanish . at zero wave number
we thus have one zero eigenvalue that corresponds to one goldstone mode , in agreement with the expectation from sec .
[ subsubsec : iii.b.1 ] .
we next calculate the gradient - squared terms .
geometric identities similar to those expressed in eqs .
( [ eqs : c.3 ] ) allow to determine the vertices .
one finds a gaussian action of the form given by eq .
( [ eq:3.15 ] ) . for @xmath338 ,
the matrix @xmath248 is given by \(k ) = ( ccc ak^2 + bk_^2 & -ick_y & -ick_x + ick_y & cq + ak^2 + bk_z^2 & i(a+b)qk_z + ick_x & -i(a+b)qk_z & cq + ak^2 + bk_z^2 ) ccc + + ( b>0 ) . [ eq : c.4 ] for @xmath252 , the matrix @xmath248 takes the form _
00(k ) & = & ( a + b/3)k^2 , + _
0+(k ) & = & _ + 0(k)^ * = i(c + bq)(k_|| - 3k_z ) + b(k_x^2 - k_y^2 ) , + _
0-(k ) & = & _ -0(k)^ * = i(c + bq)(k_x - k_y ) - b(k_x^2 + k_y^2 - 2k_z^2 ) , + _ + +
( k ) & = & _ (k ) = cq + ( a + b/3)k^2 , + _
+ -(k ) & = & _ -+(k)^ * = i(a+b/3 ) qk_|| ( b<0 ) .
[ eq : c.5 ] calculating the eigenvalues one finds eq .
( [ eq:3.16 ] ) .
in this appendix we sketch how the expressions for the single - particle and transport relaxation rates given in sec .
[ subsec : iv.b ] change in the ferromagnetic limit
. a more thorough discussion of this topic will be given elsewhere.@xcite there are three major changes compared to eqs .
( [ eq:4.4 ] ) and ( [ eq:4.7 ] ) that take place in the ferromagnetic limit .
firstly , there is no intra - stoner - band scattering .
this is because in a ferromagnet , the goldstone mode is entirely transverse with respect to the direction of the magnetization , whereas in a helimagnet this is not the case.@xcite as a result , with @xmath281 in eq .
( [ eq:4.4 ] ) on the 1-fermi surface , the resonance frequency inside the delta function under the integral will be @xmath339 .
secondly , the dimensionless vertex @xmath340 , eq.([eq:4.6b ] ) , changes to unity . technically , this is seen most easily within the formalism of ref . .
physically , it reflects the fact that for intra - stoner - band scattering the magnetic fluctuations couple to the quasi - particle density , and thus physically act akin to a chemical potential , only gradients of which contribute to scattering . for inter - stoner - band scattering , on the other hand , they physically act akin to an external magnetic field , and thus they couple without gradients .
thirdly , the form of the susceptibility @xmath286 , eq .
( [ eq:4.6c ] ) , changes .
it now reflects the ferromagnetic magnons , and for power counting purposes it can be adequately represented by [ eqs : d.1 ] ( k , u ) = ( u - _ ( k ) ) , [ eq : d.1a ] with _ ( k ) = const.k^2/^2 [ eq : d.1b ] the ferromagnon frequency . putting all of this together , and neglecting numerical prefactors , we obtain for the single - particle scattering rate due to ferromagnetic magnons [ eqs : d.2 ] = _ p
( _ 2(k + p ) - _ ( p ) ) , [ eq : d.2a ] and for the electrical transport rate & = & _ p , k ( _ 1(k ) ) + & & ( _ 2(k+p ) - _ ( p ) ) .
[ eq : d.2b ] [ eq : d.4 ] we provide some brief comments regarding these results , a more complete discussion will be given elsewhere.@xcite ( 1 ) the energy scales @xmath341 and @xmath249 that lead to the three temperature regimes shown above emerge within the bare ( stoner - level ) theory .
it is not obvious how these scales become renormalized , and one therefore has to be careful when making quantitative comparisons with experiment . in particular , within the bare theory the regime @xmath342 is not realizable since it implies @xmath343 ; this may change within a fully renormalized theory .
( 2 ) the second result in eq .
( [ eq : d.4 ] ) , @xmath344 , reproduces the result of ueda and moriya.@xcite we note , however , that this is _ not _ the true asymptotic low - temperature result ; it is valid only in a pre - asymptotic temperature window .
( 3 ) the true asymptotic behavior for @xmath345 is exponential , rather than power - law , in @xmath18 .
this is due to the stoner splitting of the conduction electrons ; this effect was neglected in ref . .
in helimagnets the situation is qualitatively different , see sec .
[ subsec : iv.b ] . | a field theory appropriate for magnets that display helical order due to the dzyaloshinsky - moriya mechanism , a class that includes mnsi and fege , is used to derive the phase diagram in a mean - field approximation .
the helical phase , the conical phase in an external magnetic field , and recent proposals for the structure of the a - phase and the non - fermi - liquid region in the paramagnetic phase are discussed .
it is shown that the orientation of the helical pitch vector along an external magnetic field within the conical phase occurs via two distinct phase transitions .
the goldstone modes that result from the long - range order in the various phases are determined , and their consequences for electronic properties , in particular the specific heat , the single - particle relaxation time , and the electrical and thermal conductivities , are derived .
various aspects of the ferromagnetic limit , and qualitative differences between the transport properties of helimagnets and ferromagnets , are also discussed . |
spatial confinement of condensed matter is known to induce a wealth of exotic crystalline structures.@xcite in essence this can be attributed to a phenomenon coined packing frustration ; an incompability between the preferred packing of particles whether atoms , molecules , or colloidal particles and the packing constraints imposed by the confining surfaces . as an illustrative example , we can consider the extensively studied system of hard spheres confined between planar hard surfaces at a close separation of about five particle diameters or less .
this is a convenient system for studies on packing frustration , because its phase diagram is determined by entropy only .
while the phase diagram of the bulk hard - sphere system is very simple,@xcite the dense packing of hard - sphere particles in narrow slits has been found to induce more than twenty novel thermodynamically stable crystalline phases , including exotic ones such as buckled and prism - like crystalline structures.@xcite in the case of spatially confined fluids , the effects of packing frustration are more elusive .
nevertheless , extensive studies on the fluid s equilibrium structure has brought into evidence this phenomenon ; confinement - induced ordering of the fluid is suppressed when the short - range order preferred by the fluid s constituent particles is incompatible with the confining surface separation ( see , e.g. , ref . for illustrative examples ) .
packing frustration also influences other properties of the confined fluid , such as a strongly suppressed dynamics because of caging effects.@xcite however , little is known to date about the underlying mechanisms of frustration in spatially confined fluids .
a stumbling block when elucidating the mechanisms of packing frustration in fluids is the hierarchy of distribution functions;@xcite a mechanistic analysis of distribution functions requires higher - order distributions as input . while density profiles ( i.e. , singlet distributions ) in inhomogeneous fluids are routinely determined today , either by theory , simulations , or experiments , structural studies are only seldom extended to the level of pair distributions.@xcite the overwhelming majority of theoretical work in the literature has been done on the singlet level where pair correlations from the homogeneous bulk fluid are used in various ways as approximations for the inhomogeneous system .
moreover , even in the cases where the pair distributions for the inhomogeneous fluid have been explicitly determined,@xcite the mechanistic analysis of ordering is hampered by the sheer amount of variables .
a conceptually simple scheme for addressing ordering mechanisms in inhomogeneous fluids is therefore much needed . in this work , we deal with the mechanisms of packing frustration in a dense hard - sphere fluid confined between planar hard surfaces by means of first - principles statistical mechanics at the pair distribution level . for this purpose
we introduce principal components of the mean force acting on a particle , and study their behavior as a function of confining slit width .
this provides a novel and conceptually simple scheme to analyze the mechanisms of ordering in inhomogeneous fluids .
in contrast to the aforementioned multitude of exotic crystalline structures induced by packing frustration , we obtain compelling evidence that even for a dense hard - sphere fluid in narrow confinement , as studied here , the ordering is in essence governed by the packing constraints at a single solid - fluid interface .
nonetheless , there are also some common features for the structures in the fluid and in the solid phases .
finally , we demonstrate how subtleties in the ordering may lead to important , nontrivial confinement effects .
the calculations in this work are done in integral equation theory for inhomogeneous fluids at pair correlation level , where the density profiles and anisotropic pair distributions are calculated self - consistently .
the only approximation made is the closure relation used for the pair correlation function of the inhomogeneous fluid .
we have adopted the percus - yevick closure , which is suitable for hard spheres .
the resulting theory , called the anisotropic percus - yevick ( apy ) approximation , has been shown to give accurate results for inhomogeneous hard sphere fluids in planar confinement.@xcite in principle , pair distribution data for confined fluids could also be obtained from particle configurations obtained by computer simulation , e.g. grand - canonical monte carlo ( gcmc ) simulations . however , even with the computing power presently available , one would need impracticably long simulations in order to obtain a reasonable statistical accuracy for the entire pair distribution , which for the present case has three independent variables .
for the confined hard sphere fluid , the pair distribution function has narrow sharp peaks ( see ref . for typical examples ) , which are particularly difficult to obtain accurately .
the alternative to use , for example , the widom insertion method to calculate the pair distribution point - wise by simulation is very inefficient for dense systems .
it should be noted , however , that in cases where direct comparison is feasible in practice , simulations and anisotropic integral equation theory are in excellent agreement in terms of pair distributions.@xcite in these cases , for a corresponding amount of pair - distribution data of essentially equal accuracy , the integral - equation approach was found to be many thousands times more efficient in cpu time than the simulations .
finally we note that other highly accurate theoretical approaches , such as fundamental measure theory ( see , e.g. , ref . for a recent review ) , have not yet been extended to the level of pair distributions in numerical applications .
within the present study , we focus on a dense hard - sphere fluid confined between two planar hard surfaces . for a schematic representation of the confinement geometry , we refer to fig .
[ fig : schematic ] .
the particle diameter is denoted by @xmath0 and the surface separation by @xmath1 .
the space available for particle centers is given by the reduced slit width , which is defined as @xmath2 .
the @xmath3 coordinate is perpendicular while the @xmath4 and @xmath5 coordinates are parallel to the confining surfaces .
the system has planar symmetry and therefore the number density profile @xmath6 only depends on the @xmath3 coordinate .
except when explicitly stated otherwise , the confined fluid is kept in equilibrium with a bulk reservoir of number density @xmath7 .
the average volume fraction of particles in the slit , @xmath8 , then varies between about 0.33 and 0.37 depending on the surface separation in the interval @xmath9.@xcite , the surface separation by _
h _ , and the reduced slit width by _
the gray region depicts the excluded volume around the left particle , which in this figure is in contact with the bottom wall .
the red arrow shows the collisional force exerted by the right particle on the left one .
the force acts in the radial direction .
[ fig : schematic ] , width=302 ] due to the planar symmetry all pair functions depend on three variables only , e.g. , the pair distribution function @xmath10 , where @xmath11 with @xmath12 denotes a distance parallel to the surfaces . in graphical representations of such functions , we let the @xmath3 axis go through the center of a particle at @xmath13 , i.e. , we select @xmath14 . then the function @xmath15 states the pair distribution function at position @xmath16 , given a particle at position @xmath17 .
likewise , @xmath18 gives the average density at @xmath19 around a particle located at @xmath17 .
we plot for clarity also negative values of @xmath20 , i.e. , in the following plots of pair functions @xmath20 is to be interpreted as a coordinate along a straight line in the @xmath21 plane through the origin . throughout this study
, we make use of integral equation theory for inhomogeneous fluids on the anisotropic pair correlation level . following refs . and
, we determine the density profiles @xmath22 and pair distribution functions @xmath15 of the confined hard - sphere fluid by solving two exact integral equations self - consistently : the lovett - mou - buff - wertheim equation , @xmath23}{dz_1}= \int c(z_1,z_2,r_{12})\frac{dn(z_2)}{dz_2 } dz_2 d{\bf r}_{12 } , \label{eq : lmbw}\ ] ] and the inhomogeneous ornstein - zernike equation , @xmath24 where @xmath25 is the total and @xmath26 the direct pair correlation function , while @xmath27 denotes the hard particle - wall potential @xmath28 as the sole approximation , we thereby make use of the percus - yevick closure for anisotropic pair correlations , @xmath29 , where @xmath30 denotes the cavity correlation function that satisfies @xmath31 and @xmath32 is the hard particle - particle interaction potential , @xmath33 this set of equations constitutes the apy theory . the confined fluid is kept in equilibrium with a bulk fluid reservoir of a given density by means of a special integration routine , in which the rate of change of the density profile for varying surface separation is given by the exact relation@xcite @xmath34 \label{eq : profile_equil}\end{gathered}\ ] ] under the condition of constant chemical potential .
here we have explicitly shown the @xmath35 dependence of the functions , which is implicit in the previous equations . for a concise review of the theory and details on the computations we refer to ref . .
for the hard - sphere fluid confined between hard planar surfaces .
( a ) data for the average volume fraction @xmath36 of particles in the slit of width @xmath37 ( reduced slit width @xmath38 ) , which is virtually at phase separation to the crystalline state for this surface separation .
the solid line depicts theoretical data within the anisotropic percus - yevick ( apy ) approximation , while the crosses show data from the grand - canonical monte carlo simulation of ref . .
( b ) theoretical data from apy approximation for a confined fluid in equilibrium with a bulk reservoir of number density @xmath7 .
the reduced slit widths are @xmath39 ( dotted line ) , @xmath40 ( solid line ) , and @xmath41 ( dashed line ) .
the average volume fraction @xmath42 is here 0.35 , 0.34 , and 0.33 , respectively .
[ fig : simul ] , width=321 ]
the theoretical approach adopted in this work has recently been shown@xcite to be in quantitative agreement with experiments at the pair distribution level for a confined hard - sphere fluid in contact with a bulk fluid of the same density , @xmath7 , as used in the current work .
both the anisotropic structure factors from pair correlations and the density profiles agree very well with the experimental data . for higher densities ,
we compare in fig .
[ fig : simul](a ) our result with the density profile obtained from gcmc simulations by mittal et al.@xcite for an average volume fraction in the slit @xmath36 and at @xmath43 . for this extreme particle density , which is virtually at phase separation to the crystalline phase at this surface separation,@xcite there are quantitative differences , but our theoretical profile agree semi - quantitatively with the simulation data . for @xmath44 and @xmath45 and at the same @xmath42 , the deviations between our profiles and the gcmc profiles by mittal et al .
are larger ( not shown ) . in the rest of this paper
we shall , however , treat cases with lower particle concentrations in the slit : @xmath42 between about 0.33 and 0.37 , which are less demanding theoretically . in ref .
we showed , for a wide range of slit widths , that our theory is in very good agreement with gcmc simulations for the confined hard sphere fluid in equilibrium with a bulk density @xmath46 , which is only slightly lower than what we consider in this work .
furthermore , our main concern in this paper are cases with surface separations about halfway between integer multiples of sphere diameters , as in fig . [
fig : simul](a ) . returning to the system in equilibrium with a bulk with density @xmath7
, we illustrate the concept of packing frustration in spatially confined fluids by presenting the number density profile @xmath6 for reduced slit widths of @xmath47 , @xmath40 , and @xmath41 in fig .
[ fig : simul](b ) .
the fluid in the narrowest slit exhibits strong ordering , as illustrated by well - developed particle layers close to each solid surface .
such ordering is observed for the hard - sphere fluid in narrow hard slits when the surface separation is close to an integer multiple of the particle diameter @xmath0 . in this specific case ,
the average volume fraction @xmath48 is about @xmath49 of the volume fraction for phase separation to the crystalline phase at this surface separation,@xcite and the `` areal '' number density near each solid surface is @xmath50 , about @xmath51 of the freezing density for the two - dimensional hard - sphere fluid.@xcite for slit widths intermediate between integer multiples of @xmath0 , the confined fluid develops into a relatively disordered fluid in the slit center , despite the confining slit being narrow enough to support ordering across the slit . in the particular case shown in fig .
[ fig : simul](b ) , we observe shoulders in the density peaks close to each surface which evolve with increasing slit width into two small ( or secondary ) density peaks in the slit center . at slightly larger slit widths ( to be investigated below ) ,
these two small peaks merge and form a fairly broad layer in the middle of the slit .
for @xmath52 there is strong ordering again ; the layers at either wall are then very sharp and the mid - layer is quite sharp ( more profiles for @xmath53 for the current case can be found in ref . and as a video in ref . ) .
such a change of ordering at the intermediate separations is usually interpreted as a signature of packing frustration , and in this paper we will address its mechanisms .
how can we understand these observations ?
the starting point for our discussion will be the pair density @xmath54 , i.e. , the density at position @xmath19 given a particle at position @xmath13 . as will become evident below , the pair densities allow us to analyze the mechanisms leading to the detailed structure of the layers in confined , inhomogeneous fluids . here
, we shall in particular investigate the mechanisms of packing frustration in dense hard - sphere fluids under spatial confinement .
[ fig : ng ] shows examples of contour plots of the pair density @xmath18 for three reduced slit widths , @xmath55 , @xmath56 , and @xmath57 , when a particle ( the `` central '' particle ) is located on the @xmath3 axis at coordinate @xmath58 .
the density profiles for these three slit widths are shown in the right hand side of the plot . in fig .
[ fig : ng](a ) there is a shoulder in the profile on either side of the midplane , while in fig .
[ fig : ng](b ) two small , but distinct , peaks have formed near the slit center . in fig .
[ fig : ng](c ) these secondary peaks have merged into one peak in the middle .
these changes in the density profile occur within a variation in surface separation of only @xmath59 . in the contour plots ,
the central particle s center is in all cases situated at a distance of @xmath60 ( about three particle radii ) from the bottom surface , @xmath61 , marked by a filled circle in the profile .
particles that form the main layer in contact with the bottom surface can then touch the central particle ; the latter is penetrating just the edge of this layer .
note that the position of the secondary maximum for the middle case , fig .
[ fig : ng](b ) , is also located at @xmath62 .
thus , it can be understood that particles forming the small secondary peak in fig .
[ fig : ng](b ) are in contact with , but barely penetrating , the main layer of particles at the bottom surface .
the particles of this secondary peak are at the same time strongly penetrating the main layer at the top surface .
the same is , however , true for the particles around the same @xmath3 coordinate ( @xmath63 ) in figs .
[ fig : ng](a ) and [ fig : ng](c ) , but with a markedly different outcome for the profile .
our task here is to understand the reason for such differences . in the contour plots of the pair density @xmath18 in fig .
[ fig : ng ] we see that in all three cases the particle density in the wedge - like section formed between the central particle and the upper wall is strongly enhanced , resulting in a local number density of up to 17 , 20 , and 24 @xmath64 , respectively , in the three cases .
this enhancement in @xmath65 relative to the density @xmath66 at the same @xmath3 coordinate is given by the pair distribution function @xmath67 , which is about 4 4.5 in the inner part of the wedge - like section for all these cases .
in the region near the bottom surface , where the central particle is in contact with the main bottom layer , there is also an enhancement in density , but to much smaller extent than at the top .
note that the density distribution @xmath65 near the bottom is very similar in all three cases . to understand why the profiles differ so much in these three cases , we investigate the mean force @xmath68 that acts on a particle with its center at position @xmath3 .
the potential of mean force , @xmath69 , is related to the density @xmath66 by @xmath70 , where @xmath71 , @xmath72 is boltzmann s constant , and @xmath73 the absolute temperature .
this implies that @xmath74 . due to the planar symmetry
, @xmath66 depends on @xmath3 only and the total force components in the @xmath4 and @xmath5 directions are zero .
the mean force @xmath75 is then directed parallel to the @xmath3 axis and we have @xmath76 , where @xmath77 .
thus , an understanding of the behavior of the profile can be obtained from an analysis of @xmath75 .
the sign of @xmath75 tells whether @xmath66 is increasing or decreasing and extremal points of @xmath66 correspond to points where @xmath75 is zero . for hard - sphere fluids , the forces exerted on a particle by the other particles in the system
are simply due to collisions . due to planar symmetry
, the density distribution in the vicinity of a particle has rotational symmetry around the @xmath3 axis through the particle center .
the average force at each point is acting in the direction normal to the sphere surface and for a particle located at @xmath78 the average force from all collisions along the sphere periphery at coordinate @xmath79 is proportional to the contact density @xmath80 , where @xmath81 is the contact value of the pair distribution function at the particle surface .
note that the force in , for example , the @xmath4 direction on one side of the periphery is cancelled by the force in the @xmath82 direction on the opposite side .
thus only the @xmath3 component of the net force on the particle contributes as expected .
using the first born - green - yvon equation one can show @xcite that @xmath83 ( the integral is over the range @xmath84 where @xmath85 is defined ) .
the role of the factor @xmath86 is to project out the @xmath3 component of the contact force .
( this line of reasoning is readily extended to systems exhibiting soft interaction potentials , such as lennard - jones fluids or electrolytes ; in such cases , however , one also needs to include the interactions with the walls and all other particles in the system , see e.g. refs . and . )
let us now return to the intriguing formation of secondary density maxima for @xmath87 . for this purpose , we present in fig . [ fig : ng ] the @xmath3 component of the contact forces acting on the particle .
they are represented by the arrows along the sphere periphery . in these plots , there are two major contributions to the net force acting on the particle , namely the repulsive forces exerted by the particle layers close to each confining wall . for @xmath88 and
the chosen position of the central particle in fig .
[ fig : ng](b ) , @xmath61 , these force contributions cancel each other : the sum of the arrows ( with signs ) is zero and hence @xmath89 at this @xmath3 coordinate , as shown to the right in the figure .
it is the subtle interplay between these forces for neighboring @xmath78 values which leads to the secondary density maximum .
the situation is , however , markedly different for @xmath90 and @xmath57 .
while the total force exerted by the particles in the main layer at the bottom surface is practically equal for all three cases , the magnitude of the force exerted by the particles in the main layer at the upper surface varies strongly with @xmath35 .
this variation is partly due to the different magnitude of the contact densities in the wedge - like region mentioned above and partly due to the change in angle between the normal vector to the sphere surface there and the @xmath3 axis . recall that the contact force acts along this normal vector , so the @xmath3 component is dependent on this angle . for @xmath90 , fig .
[ fig : ng](a ) , the @xmath3 component of the contact force from the upper layer is smaller than for @xmath88 .
the sum of the arrows is then positive , i.e. the total average force is directed towards the upper wall and hence @xmath91 at this @xmath3 coordinate . for @xmath92 , fig .
[ fig : ng](c ) , this @xmath3 component is larger compared to @xmath88 , thereby pushing the central particle towards the slit center .
hence @xmath93 at this @xmath3 coordinate . . the force contributions acting in positive ( @xmath94 , full curve ) and negative ( @xmath95 , dashed curve )
@xmath3 directions are presented separately as functions of particle position @xmath78 across the confining slit .
data are shown for reduced slit widths @xmath90 , @xmath56 , and @xmath57 . the values of the forces for a particle at the @xmath78 coordinates in fig .
[ fig : ng ] are shown by filled circles .
the dashed vertical line denotes the slit center , while the solid vertical line on the right - hand side indicates the upper limit for possible @xmath78 coordinates of the particle in the slit . for comparison of all three cases ,
@xmath94 is also shown for @xmath90 ( blue dots ) and @xmath56 ( black dots ) in the bottom panel .
[ fig : force ] , width=302 ] in order to gain more insight into the formation of the secondary maxima , we present in fig .
[ fig : force ] the net force acting on a particle for all positions @xmath78 in the same three cases as discussed above , @xmath90 , @xmath56 , and @xmath57 . to facilitate the interpretation , the principal force contributions acting in positive ( denoted @xmath94 ) and negative ( @xmath95 )
directions are shown separately .
the total net force is @xmath96 , where @xmath94 originates from collisions on the lower half of the sphere surface and @xmath95 on the upper half ( @xmath94 and @xmath95 correspond to the absolute values of the sums of arrows in respective hemisphere in fig .
[ fig : ng ] ) . in fig .
[ fig : force ] the red ( solid ) and black ( dashed ) curves are each other s mirror images with respect to the vertical dashed axis at @xmath97 , which shows the location of the slit center .
since the variation in @xmath94 ( and @xmath95 ) is very similar for all slit widths in fig .
[ fig : force ] , the following discussion will hold for all three cases . for @xmath98
we have @xmath99 , because no spheres can collide from below since the confining surface precludes them from being there ( cf .
[ fig : schematic ] ) . with increasing @xmath78 we observe a monotonically increasing @xmath94 , which can be attributed both to the increasing area exposed to collisions on the lower half of the sphere surface and the decrease in angle of the sphere normal there relative to the @xmath3 axis . with further increase in @xmath78
, we eventually observe a decrease in the exerted force induced by a decrease in contact density @xmath80 .
around @xmath100 we observe a sudden onset of a rapid decrease for @xmath94 .
this is a consequence of a rapid decrease in contact density , that occurs when the particle at @xmath78 loses contact with the dense particle layer at the bottom wall . for even larger @xmath78 , where the particle is close to the top surface ,
collisions with particles in the slit center around the entire lower half of the sphere surface become important so that @xmath94 increases again .
the three red curves are compared in the bottom panel , where @xmath94 from the first two panels ( @xmath90 and @xmath56 ) are shown as dotted curves .
we see that the curves are nearly identical apart from in a small region to the far right .
the analogous statement is true , of course , for the black dashed curves .
thus , apart from small @xmath78 intervals to the extreme left and right , the behavior of @xmath96 for the different slit widths can be understood in terms of a horizontal shift of the red and the black curves relative to each other . the formation of secondary density maxima
can then be explained from the resulting balance of the force contributions . for @xmath101 and @xmath102 [ i.e. , the right half of fig .
[ fig : force](b ) ] , the two curves intersect at two points where the forces cancel each other and where @xmath89 . the intersection marked by the filled circle gives a local maximum of @xmath6 and the next one to the right gives a minimum . together with the minimum at the slit center , @xmath97 , where the curves also cross each other , these features give rise to the secondary peak of the density profile as we have seen in fig . [
fig : ng](b ) .
this subtle balance of forces , and hence the formation of secondary maxima , is only observed in a narrow range of slit widths , as evidenced by the force profiles for @xmath90 and @xmath57 . in the latter case
, the intersection at @xmath97 corresponds to a local maximum and the other one to a minimum .
together they give one peak in the middle as seen in fig .
[ fig : ng](c ) .
the formation of secondary maxima is for @xmath102 accordingly a consequence of two phenomena : ( i ) the rapid decrease of @xmath94 followed by the subsequent increase of @xmath94 and ( ii ) the monotonic decrease of @xmath95 in the same region .
together these effects lead to the force curves intersecting twice in the manner they do for @xmath101 .
the rapid decrease of @xmath94 is , as we have seen , due to the loss of contact of the particle with the well - developed bottom layer , while the monotonic decrease of @xmath95 occurs when the particle approaches the top surface . , but for reduced slit widths @xmath103 , @xmath104 , and @xmath105 . [ fig : force2 ] , width=302 ] for comparison , we present in fig . [
fig : force2 ] the principal force components for a set of larger slit widths : @xmath106 , @xmath104 , and @xmath105 .
there are no secondary maxima in this case .
instead we observe for @xmath107 a broad region in the center of the slit where @xmath94 and @xmath95 virtually cancel each other and where , as a consequence , @xmath108 .
hence , this observation implies an essentially constant @xmath66 in the slit center , as can be seen in the third full curve of fig .
[ fig : n_z ] , where density profiles are shown for various cases . for the confined hard - sphere fluid .
the reduced slit widths are @xmath101 ( offset vertically by 3.0@xmath64 ) , @xmath109 ( offset by 2.0@xmath64 ) , and @xmath104 ( offset by 1.0@xmath64 ) .
the systems are otherwise the same as in fig .
[ fig : simul](b ) .
the solid and dotted lines depict results based on the full apy theory and the superposition approximation , respectively . the density profile at a single solid - fluid interface ( @xmath110 ) is also shown for comparison .
[ fig : n_z ] , width=321 ] the course of events shown in fig .
[ fig : force2 ] when we increase @xmath35 from @xmath111 to @xmath105 implies the formation of a layer at the slit center .
the crossing of the principal force curves in fig .
[ fig : force2](a ) at the slit center , @xmath97 , corresponds to a density minimum , while that in fig .
[ fig : force2](c ) corresponds to a density maximum . note that for @xmath112 there are four layers in the slit ( two very sharp ones at the walls and two less sharp on either side of the slit center ) and for @xmath113 there are five layers .
the fifth layer that forms in the middle for the intermediate separations arises via the broad flattening of the density profile in the middle , and signals the packing frustration in this case . the data of figs .
[ fig : force ] and [ fig : force2 ] indicate a qualitative difference in @xmath6 for @xmath87 and @xmath114 . in the transition from @xmath115 particle layers ,
the third layer is formed via the occurrence of secondary layers close to each surface , which merge to form a central layer with increasing @xmath35 .
this contrasts the transitions from @xmath116 particle layers just discussed , where the new particle layer forms directly in the slit center .
the secondary peaks for @xmath87 are also evident in qualitatively different anisotropic structure factors @xmath117 for @xmath118 and @xmath119 presented in our previous work , ref . .
@xmath117 for confined fluids is governed by an ensemble average of the anisotropic pair density correlations @xmath120 ( see ref . for more slit widths ) . in order to address these differences in @xmath6 with @xmath35 , we will in the following
analyze further the principal force component @xmath94 . in both figs .
[ fig : force ] and [ fig : force2 ] , the principal force components @xmath94 ( and @xmath95 ) for different @xmath35 nearly coincide for most @xmath3 values . in order to investigate this further
, we plot @xmath94 for a wider set of slit widths , @xmath121 , in fig .
[ fig : f_up](a ) . indeed , apart from rather small deviations at large @xmath3 , all data fall on a _ master curve _ given by @xmath94 for @xmath110 , i.e. , the force component for the single solid - fluid interface [ the former curves are also shown separated in fig .
[ fig : f_up](b ) ] .
although not shown here , we have verified that this observation holds reasonably well for @xmath122 , implying the same ordering mechanism irrespective of slit width . for the hard - sphere fluid between planar hard surfaces .
the reduced separations are @xmath123 , @xmath124 , @xmath119 , @xmath105 , and @xmath125 .
the systems are otherwise the same as in fig .
[ fig : simul](b ) .
( a ) @xmath94 for the confined fluids ( black lines ) and for a single solid - fluid interface ( @xmath126 , red line ) .
( b ) @xmath94 for the confined fluids ( each offset vertically by @xmath127 units for clarity ) , obtained via the full apy theory [ solid lines , same as in ( a ) ] and the superposition approximation ( dotted lines ) .
( c ) force components of the superposition approximation , @xmath128 and @xmath129 , for different reduced slit widths ( the latter curves are vertically offset by @xmath130 for clarity ) .
@xmath128 is the same as the red curve in ( a ) .
[ fig : f_up ] , width=321 ] in order to gain further insight into the ordering mechanism , we have determined density profiles obtained in a simple superposition approximation.@xcite within this approximation , the potential of mean force @xmath69 in the slit is calculated as the sum of the corresponding potentials from two single hard surfaces , i.e. @xmath131 , where @xmath132 denotes the potential of mean force for the fluid at a single solid - fluid interface in contact with a bulk fluid of density @xmath133 . this implies the superposition for the mean force : @xmath134 .
since the density profile is given by @xmath135 $ ] the superposition approximation implies @xmath136 where we have explicitly shown that the density profile for the slit , @xmath137 , depends on @xmath35 , and where superscript sp indicates `` superposition '' and @xmath138 is the density profile outside a single surface . in fig .
[ fig : n_z ] we compare @xmath6 for reduced slit widths of @xmath88 , @xmath109 , and @xmath104 obtained via the full theory ( solid lines ) and the superposition approximation thus obtained ( dotted lines ) .
note that there are density peaks at @xmath139 for all three slit widths and that they approximately coincide with the location of a density peak for the single solid - fluid interface ( also shown in fig .
[ fig : n_z ] ) .
this implies that the density peak at @xmath139 is strongly correlated with the bottom solid surface .
although the profiles obtained via the superposition approximation deviate quantitatively from those of the full theory , especially for narrow slit widths , the qualitative agreement implies that the main features of @xmath6 the density peaks and shoulders of fig .
[ fig : n_z ] are rather uncomplicated confinement effects . to substantiate this conclusion , we present in fig . [ fig : f_up](b ) the principal force components @xmath94 for @xmath140 , obtained both using the full theory and the superposition approximation .
the agreement is equally good as for the density profile of the @xmath141 case in fig .
[ fig : n_z ] .
a significant point is now that the superposition allows us to separate the contributions to @xmath94 from each surface in a simple manner , that will provide insights into what happens during confinement .
as shown in appendix [ sec : appendixa ] , @xmath94 can be decomposed in this approximation into two components : a major contribution from the lower surface , @xmath142 , and a correction due to the presence of the upper surface , @xmath143 . the former is the same as the average force component for the single solid - fluid interface plotted in fig .
[ fig : f_up](a ) ( denoted as `` master curve '' above ) .
we have @xmath144 where @xmath145 , see eq .
( [ eq : fup_splita ] ) . here , @xmath146 is the average force for the case of a single solid - fluid interface ( u ) and @xmath147 is the force that acts on one side of a hard sphere ( i.e. on one half ) in the bulk fluid .
note that in @xmath148 it is only @xmath146 that depends on @xmath35 . and @xmath143 in the superposition approximation .
each arrow represents a force that acts on the entire red half of the sphere ( the location of the arrow has no significance in this sketch ) .
the lower wall is shown in black and the upper wall is to be placed on the location indicated by the striped rectangle . the dashed line that connects each arrow to the respective surface indicates from which wall the influence originates .
[ fig : sketch_fsp ] , width=245 ] in fig .
[ fig : f_up](c ) we show @xmath142 and @xmath143 for the same surface separations as before .
the @xmath35 dependence of the latter is simply a parallel displacement along @xmath3 . when @xmath142 and @xmath143 are added we obtain the dotted curves in fig .
[ fig : f_up](b ) . thus the differences between each black curve and the red curve in fig .
[ fig : f_up](a ) is essentially contained in the contribution @xmath143 from the upper surface ( for smaller surface separations there will remain a minor difference as indicated by the small deviations for the superposition approximation in fig .
[ fig : n_z ] ) . to see in more detail what this means
, we have in fig .
[ fig : sketch_fsp ] shown schematically how these force contributions act on a sphere . in the presence of only one solid - fluid interface ( l ) , the total force in the direction away from the surface ( upwards ) is @xmath149 , i.e. , the force on the bottom half of the sphere shown as red in the figure .
let us now place the second surface ( u ) some distance from the other , at the location indicated in the figure .
the change in the upwards force due to this second surface is given by @xmath150 in the superposition approximation .
note that the former force , @xmath142 , acts on the hemisphere that is facing the surface l , while the latter , @xmath143 , is a force that acts on the hemisphere _
away from _ the corresponding surface u and in the direction _ towards _ this surface .
if the lower wall were not present when we place the upper wall at the indicated position , the initial state would be a homogeneous bulk fluid and the final state a single solid - fluid interface ( u ) in contact with the bulk .
in this situation @xmath143 equals the actual change in the average force on the red hemisphere . in eq .
( [ eq : fup_split ] ) we have adopted this value as an approximation for the corresponding change when placing the upper wall in the presence of the lower one , i.e. , when the initial state is an inhomogeneous fluid in contact with the lower surface and the final state is a fluid simultaneously affected by both surfaces . since
this approximation obviously is very good , it follows that the inhomogeneity due to one surface has only a small influence on the effects from the other surface throughout the entire slit .
we saw in section [ sec : principal_comp ] that the seemingly complicated changes in structure as the surface separation varies around half - integer @xmath0 values of @xmath35 ( i.e. , @xmath151\sigma$ ] with @xmath152 integer ) , can be mainly explained by a parallel displacement of upward and downwards force curves along the @xmath3 direction .
there was , however , some variation in these force curves near one of the surfaces ( the upper surface for the upward forces and the lower surface for the downwards forces ) that remained unexplained there . in the current section
we have seen that this variation too can be mainly explained by a parallel displacement in this case a displacement of the contributions to @xmath94 ( or @xmath95 ) due to each surface as seen in fig .
[ fig : f_up](c ) . to summarize our results in this section
we make two important conclusions : first , by considering the mean force due to one surface ( here the lower one ) and by treating the influence from the other ( upper ) surface as a correction @xmath129 according the the superposition approximation , one obtains nearly quantitative agreement with the full theory .
our approach of defining principal components of the mean force thereby provides a means to understand the contributions of each confining surface .
second , the principal force components obtained within the full theory and the superposition approximation are virtually in quantitative agreement for @xmath153 . for narrower slit widths ( down to @xmath154 ) ,
quantitative discrepancies become more pronounced .
these quantitative differences , which will be discussed in the next subsection , are nontrivial confinement effects .
nevertheless , the semi - quantitative agreement in the whole range of slit widths , down to @xmath154 , further strengthens the notion that ordering of confined hard - sphere fluids can , to a good approximation , be explained as a single - wall phenomenon .
in essence , the fluid conforms locally with only one of the confining surfaces at a time . in some local regions it will thereby conform to one surface and in other regions to the other surface
regions that are continuously changing ( recall that the distributions we calculate are time averages of the various possible structures ) .
we emphasize that this reasoning holds for all slit widths , irrespective of whether @xmath35 is close to an integer or a half - integer multiple of the particle diameter ( cf .
[ fig : f_up ] ) .
in other words , from a mechanistic point of view there is little difference between ordering in frustrated and more ordered confined hard - sphere fluids . in the latter case ,
the local ordering near one surface essentially agrees with the local ordering at the other one , whereby for the density profiles there appear only small mutual effects of the ordering from both surfaces beyond what is given by superposition .
an interesting similarity between the structures observed in the fluid and solid phases should be mentioned . in some of the exotic crystalline structures observed under confinement most notably the prism - like structures @xcite the particles locally conform with one of the solid surfaces .
this is reminiscent of the situation in the fluid phase discussed above , although in the latter case the structures are less ordered and constantly changing locally . in particular , the adaptive prism phase @xmath155 found in ref .
would yield an average density profile with secondary peaks on either side of the midplane , similar of those shown in fig .
[ fig : ng](b ) but much sharper .
the fact that the superposition approximation works surprisingly well for these rather large densities and gives a large part of the effects of confinement , means that it is simple to obtain good estimates of the density profiles for a confined fluid given an accurate density profile for a single solid - fluid interface . to obtain
the latter is , however , computationally nontrivial and requires fairly advanced theories .
furthermore , as we shall see below , not all important properties of the confined fluid can be explained by superposition . and ( b ) contact density @xmath156 of hard spheres between two hard planar surfaces as functions of reduced surface separation .
the system is in equilibrium with a bulk fluid of density @xmath7 [ same as in fig .
[ fig : simul](b ) ] .
data are shown for both the full apy theory ( solid lines ) and the superposition approximation ( dashed lines ) .
[ fig : contact ] , width=302 ] we have shown that the density profile @xmath6 of confined hard - sphere fluids is , to a large extent , determined by packing constraints at a single solid - fluid interface . in this respect , the ordering is a trivial confinement effect .
however , subtle deviations in @xmath6 do remain in the superposition approximation , and these may lead to important , nontrivial confinement effects .
the two most prominent nontrivial effects of confinement in , for example , fig .
[ fig : n_z ] are the slit width dependence of the contact density at the walls , @xmath157 , and the total number of particles per unit area in the slit @xmath158 , which is a fundamental quantity for many properties of the confined fluid . in the following
, we will discuss these two and related quantities in more detail . in fig .
[ fig : contact](a ) , we present the excess adsorption @xmath159 dz$ ] of particles in the slit as a function of reduced slit width @xmath35 , determined via both the full theory and the superposition approximation .
the discrepancy between the two theoretical schemes is striking ; since @xmath160 is an integrated quantity , minor systematic deviations in @xmath6 accumulate to a large effect in the total number of particles .
the superposition approximation gives , for example , in the interval @xmath44 to @xmath45 an estimate of @xmath161 that is wrong by a factor that varies between 1.36 and 0.84 .
we note that , e.g. , dynamic quantities such as diffusion coefficients@xcite and relaxation times@xcite in simple confined fluids have been found to scale with particle packing , as quantified by the excess entropy .
consequently , a systematic error in the packing of particles ( especially for very narrow confinement ) , as evidenced by systematic quantitative differences in the number density @xmath162 and an ensuing large discrepancy in @xmath163 between the full theory and the superposition approximation , will have a substantial impact on many properties of the confined fluid obtained theoretically .
[ fig : contact](b ) shows the contact density @xmath156 as a function of @xmath35 , again obtained both via the full theory and the superposition approximation .
this is an important quantity , because it yields the pressure between the walls , @xmath164 , according to the contact theorem .
consequently , @xmath157 is related to the net pressure acting on the confining surfaces , @xmath165 with @xmath166 denoting the bulk pressure , and hence to the extensively studied oscillatory surface forces.@xcite while the superposition approximation explains reasonably well the magnitude of @xmath157 , there is a nontrivial systematic phase shift with respect to @xmath35 of about @xmath167 .
this effect has been observed by one of us ( s.s . )
already earlier,@xcite and in the following we will provide a mechanistic explanation of the phenomenon .
a similar phase shift can also be seen in @xmath168 , fig .
[ fig : contact](a ) . in the superposition approximation , eq .
( [ eq : n(z)sp ] ) yields the contact density for the wall at @xmath169 as @xmath170 thus , the contact density for a reduced slit width @xmath35 is in this approximation proportional to the density at @xmath171 outside a single surface . to analyze the @xmath35 dependence
further we will need the following equation that is equivalent to eq .
( [ eq : lmbw ] ) , @xmath172}{dz_1}= \\ -\beta \int n(z_2)h(z_1,z_2,r_{12 } ) \frac{dv(z_2)}{dz_2 } dz_2 d{\bf r}_{12}. \label{eq : lmbw - h}\end{gathered}\ ] ] [ the two equations can be transformed into each other by the ornstein - zernike equation ( [ eq : oz ] ) . ] for a single hard wall - fluid interface located at @xmath169 , eq .
( [ eq : lmbw - h ] ) yields @xmath173 where @xmath174 is the total pair correlation function for the fluid outside the single surface . by inserting @xmath175 , this equation together with eq .
( [ eq : ncontact_sp ] ) imply that @xmath176 for the exact case , the corresponding equation can be obtained from eq .
( [ eq : profile_equil ] ) , which yields @xmath177 ^ 2 \int h(z_1,0,r_{12 } ) d{\bf r}_{12 } \biggl \vert_{z_1=l}. \label{eq : contact_exact}\ ] ] apart from the factors in front of the integral , we see that the main difference is that in the superposition approximation the total pair correlation function for a single wall is evaluated at coordinate @xmath175 outside the wall , while for the exact case the correlation function for the fluid in the slit is evaluated at the opposite surface ( also at @xmath175 ) . the oscillatory behavior of the contact density as a function of @xmath35 implies that its derivative changes sign with the same periodicity .
since the prefactors are positive , the phase shift for @xmath178 relative to @xmath157 must originate from the integrals . in fig .
[ fig : h ] we have plotted the total pair correlation function @xmath179 in the slit when the central particle is in contact with the lower surface ( i.e. , at coordinate 0 ) for the cases @xmath180 , @xmath181 , and @xmath105 together with the corresponding function for a single hard wall - fluid interface .
the first impression is a striking similarity of these plots , despite that there is an upper surface present in the first three cases .
there are only small differences in the entire slit compared to the single surface case for the corresponding @xmath79 values . when looking closely , one can , however , see some systematic differences in the @xmath182 function induced by the presence of the upper surface .
most importantly , we will investigate @xmath182 for @xmath175 , which occurs in the integral in eq .
( [ eq : contact_exact ] ) , and compare this with the values at the same @xmath79 coordinates for the single surface case , occurring in eq .
( [ eq : contact_sp ] ) .
these @xmath79 values are marked with red arrows in the left hand side of figs .
[ fig : h](a)(c ) and with red lines in fig .
[ fig : h](d ) . as function of @xmath20 for the systems in fig .
[ fig : h](a)-(c ) with reduced surface separations @xmath183 , @xmath184 , and @xmath185 . the data are obtained via full apy theory ( solid line ) , superposition approximation ( blue dotted line ) , and shifted superposition approximation ( @xmath186 , red dotted line ) . in the latter two cases ,
@xmath187 is plotted for the appropriate @xmath79 values ( see text ) .
note the different scales on the @xmath5 axis in the subplots .
the curves go to zero at @xmath188 because of the factor @xmath20 .
[ fig : hr ] , width=302 ] fig .
[ fig : hr ] shows @xmath189 with @xmath190 for the cases in figs .
[ fig : h](a)(c ) and these curves are compared to @xmath187 for the same @xmath79 coordinates ( shown as blue dotted lines in the figure ) .
the factor @xmath20 is included so the areas under the curves in fig . [
fig : hr ] are proportional to the values of the integrals of eqs .
( [ eq : contact_sp ] ) and ( [ eq : contact_exact ] ) ; this factor originates from the area differential @xmath191 .
the @xmath35 values in figs .
[ fig : h ] and [ fig : hr ] are selected such that we cover cases where @xmath192 and @xmath193 in fig .
[ fig : contact](b ) are negative ( @xmath180 ) and positive ( @xmath194 ) .
there is also one case ( @xmath195 ) with @xmath196 .
these signs can be verified by inspection of the areas under the curves in fig .
[ fig : hr ] ( the contributions around @xmath188 are most important for the sign ; there are substantial cancellations in the tail region due to the oscillations ) .
we can see in the figure that the full curves and the blue dotted ones do not agree , which means that the values of the integrals and hence of @xmath192 are different , as expected . if we instead plot the values of @xmath187 for @xmath197 ( red dotted lines in the figure ) we obtain better agreement
. thus the presence of the upper surface makes @xmath179 `` compressed '' in the @xmath3 direction by about @xmath167 compared to @xmath198 .
this compression gives rise to the phase shift observed in fig .
[ fig : contact ] .
there are also some other small differences between @xmath182 and @xmath174 and , in addition , there are different prefactors in eqs .
( [ eq : contact_sp ] ) and ( [ eq : contact_exact ] ) .
this gives the remaining differences in @xmath199 and @xmath200 seen in fig .
[ fig : contact](b ) .
the nontrivial confinement effects are accordingly due to rather delicate changes in the pair distribution function @xmath15 due to the presence of a second solid surface .
the packing of particles in the slit around each individual particle is described by the pair density @xmath18 and the changes in @xmath65 can be large , even for small variations in @xmath67 , in regions where the density profile @xmath66 is large .
conversely , since there are large variations in the density profiles with surface separation , the packing is strongly altered even when the change in @xmath67 is small .
the self - consistent calculation of density profiles and anisotropic pair distribution functions , as provided by integral equation theories at the pair correlation level ( like the apy theory used in this paper ) , gives efficient tools for the investigation of the structure of inhomogeneous fluids in confinement .
this is exemplified in this paper by a detailed examination of the mechanism behind the packing frustration for a dense hard - sphere fluid confined between planar hard walls at short separations .
when the width of the slit between the walls is close to an integer multiple of sphere diameters , the layer structure is optimal and the number density profile @xmath6 between the walls has sharp peaks . for slit widths near half - integer multiples of sphere diameters ( @xmath151\sigma$ ] with @xmath152 integer ) ,
the layer structure is much weaker and the packing frustration is large .
the density profile shows considerable intricacy when the slit width is varied around these latter values .
for example , when the reduced slit width @xmath35 is increased from @xmath201 , there appear secondary density peaks close to the main peaks at each wall .
these secondary peaks merge into a single peak at the slit center when @xmath35 approaches @xmath45 . the mechanism behind these and other structural changes have been investigated in this paper , using the tools provided by the anisotropic pair distribution function theory .
the number density profile is determined by the mean force @xmath68 on the particles in the slit via the relationship @xmath202 .
for the hard - sphere fluid the mean force , which acts on a particle located at @xmath3 , originates from collisions by other particles at the surface of the former
. the average collisional force on the sphere periphery is proportional to the contact density there , which varies around the surface since the fluid is inhomogeneous .
the sum of the average collisional forces constitutes the mean force @xmath75 and since we have access to the pair distribution , and thereby the contact density at the sphere surface , we can investigate the origin of any variations in @xmath75 and thereby in @xmath66 . of particular interest here are the variations when the slit width is changed . by introducing the two principal components @xmath94 and @xmath95 of @xmath75 , each of which is the sum of the average collisional forces on the particle hemisphere facing one of the walls , we extract sufficient information from the pair distributions to obtain a lucid description of the causes for the structural changes due to varying degree of packing frustration .
we show that most features of the structural changes , including the appearance and merging of the secondary peaks mentioned above , can be explained by a simple parallel displacement of the @xmath94 and @xmath95 curves when the slit width is varied around half - integer @xmath0 values .
the underlying reasons for this simple behavior is revealed via a detailed investigation of the pair distribution , that gives information about how the contact densities around the sphere periphery varies for different positions @xmath3 of a particle in the slit .
it is found that the components @xmath94 and @xmath95 , and thereby the ordering of the fluid , are essentially governed by the packing conditions at each single solid - fluid interface .
the fluid in the slit thereby conforms locally with only one of the confining surfaces at a time . in some local regions it will conform to one surface and in other regions to the other surface regions that are constantly changing ( the calculated distributions are averages of the various possible structures ) .
this picture holds for all slit widths , irrespective of whether @xmath35 is close to an integer or a half - integer multiple of the particle diameter . as a consequence of these local packing conditions , the force components @xmath94 and @xmath95 , and thereby the total mean force @xmath203 acting on a particle in the slit ,
can to a surprisingly good approximation be written as a superposition of contributions due to the presence of each individual solid - fluid interface at the walls .
when the slit width is varied , this superposition can be expressed in terms of a parallel displacement of force curves due to either surface .
there are , however , some important properties of the inhomogeneous fluid that can not be described by a simple superposition , but are instead determined by nontrivial confinement effects . in this paper , we exemplify such quantities by the number of particles per unit area in the slit @xmath161 , the excess adsorption @xmath160 , the contact density of the fluid at the wall surfaces @xmath204 , and the net interaction pressure between the walls @xmath205 . in the superposition approximation , @xmath161 and @xmath160 disagree to a large extent compared to the accurate values , while @xmath204 and @xmath205 are mainly off by a phase shift in their oscillations . the analysis show that these nontrivial confinement effects are due to rather delicate changes in the anisotropic pair distribution function @xmath15 when the wall separation is changed .
we thank tom truskett for providing the simulation data in fig . [
fig : simul](a ) . k.n . and r.k . acknowledge support from the swedish research council ( grant nos .
621 - 2012 - 3897 and 621 - 2009 - 2908 , respectively ) .
the computations were supported by the swedish national infrastructure for computing ( snic 001 - 09 - 152 ) via pdc .
for a hard sphere fluid in the slit between two hard walls , the force on , for example , the lower hemisphere of a hard sphere , @xmath94 , can in the superposition approximation be divided into contributions due to either wall surface .
the contribution @xmath142 from the lower surface is given by [ cf .
( [ eq : bgy ] ) ] @xmath206 where @xmath207 is the contact value of the pair distribution for the fluid outside a single surface .
likewise , the contribution @xmath146 from the upper surface is given by @xmath208 in the total @xmath148 there is a further contribution . from eq .
( [ eq : n(z)sp ] ) we see that the total @xmath209 is equal to the derivative of @xmath210 .
while the last term gives zero for @xmath209 , i.e. , the mean force in bulk is zero , this is not the case for @xmath211 .
the mean force on one half of the sphere surface in bulk , @xmath147 , is non - zero ; it is only the sum of the forces on both halves that are zero .
thus we have @xmath212 with @xmath213 , where @xmath214 is the contact value for the pair distribution in bulk .
when @xmath215 , the presence of the last term makes @xmath148 go to the single surface force @xmath142 , as it should in this limit .
34ifxundefined [ 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 @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed .
( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ , ( , , ) p. in @noop _ _ , ( , , ) p. @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed .
( , , ) | packing frustration for confined fluids , i.e. , the incompability between the preferred packing of the fluid particles and the packing constraints imposed by the confining surfaces , is studied for a dense hard - sphere fluid confined between planar hard surfaces at short separations .
the detailed mechanism for the frustration is investigated via an analysis of the anisotropic pair distributions of the confined fluid , as obtained from integral equation theory for inhomogeneous fluids at pair correlation level within the anisotropic percus - yevick approximation . by examining the mean forces that arise from interparticle collisions around the periphery of each particle in the slit
, we calculate the principal components of the mean force for the density profile each component being the sum of collisional forces on a particle s hemisphere facing either surface .
the variations of these components with the slit width give rise to rather intricate changes in the layer structure between the surfaces , but , as shown in this paper , the basis of these variations can be easily understood qualitatively and often also semi - quantitatively .
it is found that the ordering of the fluid is in essence governed locally by the packing constraints at each single solid - fluid interface .
a simple superposition of forces due to the presence of each surface gives surprisingly good estimates of the density profiles , but there remain nontrivial confinement effects that can not be explained by superposition , most notably the magnitude of the excess adsorption of particles in the slit relative to bulk .
published as j. chem .
phys . * 141 * , 094501 ( 2014 ) . |
multifractals appear in a wide range of situations like energy dissipation in turbulent flows @xcite , electronic eigenstates at metal insulator transition @xcite , fluctuations in finance @xcite , dynamics of human heartbeat @xcite and many others .
recently , multifractal behavior of the first passage time has been studied in the transport of diffusing particles on a @xmath7 sierpinski gasket with absorbing and reflecting barriers @xcite .
our interest is to investigate the multifractal properties associated with the collision spectrum of ballistic trajectories of point particles interacting with irregular surfaces .
the emergence of the concept of fractal geometry has provided an efficient tool to model the influence of strong geometrical irregularity on various physical and chemical processes @xcite .
ballistic trajectories in systems with irregular geometry are representative of several physical situations .
for example , they correspond to the path of light rays in irregular structures when diffraction phenomena can be neglected .
they also represent the path of atoms or molecules in confined vessels at sufficiently low pressure , @xmath8 ; in the knudsen diffusion regime @xcite .
the interaction of atoms or molecules with irregular surfaces plays a major role in heterogeneous catalysis @xcite where porous catalysts have often very irregular surfaces down to the molecular scale . due to the geometrical irregularity ,
the small scale structure may confine reactants and increase the interaction with the surface .
xenon nuclear magnetic resonance is used to probe porous structures through the interaction of rare gas atoms with the solid surfaces @xcite .
the properties of ballistic trajectories in @xmath7 pre - fractal surfaces has already been studied in connection with the chaotic aspects , possibility of `` open '' black body , knudsen diffusivity , mean interaction , and catalytic efficiency by santra et al@xcite .
coppens and co - workers have also investigated the influence of the @xmath9 fractal surface morphology on knudsen diffusion in nanoporous media @xcite .
multifractality has already been observed in the molecular diffusion on random fractal structures with time - dependent random potential @xcite .
the interest here is to explore the multifractal aspects of the number of collisions made by @xmath7 ballistic trajectories within irregular surfaces in the knudsen diffusion regime .
the spatial distribution of the collision numbers of these ballistic trajectories over the slit source is found to be multifractal .
the macro - irregularity of the surface is modeled by a quadratic koch curve of fractal dimension @xmath10 .
the pre - fractal generator is shown in fig.[demo1 ] @xmath11 .
particles are launched from a slit of unit length , placed at an angle @xmath12 with the zeroth order of the pre - fractal .
the particle trajectories due to specular collisions within the third ( @xmath13 ) generation of the pre - fractal are shown in fig.[demo1 ] @xmath14 .
solid lines with an arrow at the end represent the trajectories .
the slit source is represented by a thick line extended from @xmath15 to @xmath16 placed at @xmath17 with the zeroth stage of the pre - fractal surface , a horizontal line .
a multifractal measure is in general related to the distribution of a physical quantity on a geometrical support @xcite . here , the geometrical support is the slit source from which the trajectories are launched and the multifractal measure @xmath1 is the distribution of the relative collision numbers , ratio of the number of collisions made by a particle to the total number of collisions made by all the particles , over the slit .
the particles leave the slit normally .
each trajectory is followed as long as they are within the pre - fractal surface .
a particle trajectory is described by the successive collisions with the surface .
the collision points are determined by solving the linear equations describing the particle trajectory and the elements of the pre - fractal surface . the trajectories in fig.[demo1]@xmath14
are generated using this algorithm .
for the pre - fractal @xmath18th generation , the slit is divided into @xmath19 equal segments of length @xmath20 each .
a particle is placed at the center of each small part .
there are then @xmath19 particles launched from the slit .
the number of collisions @xmath21 made by a particle initiated from the position @xmath22 is counted and the relative collision number @xmath1 , ratio of @xmath21 to the total number of collisions @xmath23 made by all the particles , is then calculated .
the value @xmath1 is assigned to the slit segment positioned at @xmath22 . generally , particle reflections with surfaces at finite temperatures are non specular and the angle of reflection , or re - emission @xmath24 is different from the angle of incidence @xmath25 . due to the interaction with the surface , the angle of reflection
has a distribution around the angle of incidence .
also the reflection angle could be arbitrary due to the micro - roughness of the surface .
non - specular reflection for atoms or molecules may come from the complex nature of the particle - wall interaction @xcite . at the time of hitting the surface
, the particle may stay adsorbed for some time and then may desorb , exchanging momentum with the surface .
for this reason , the particle trajectories as a whole have a random character . in order to mimic the random character of each collision , the angle of reflection @xmath24
is supposed to be related to the angle of incidence @xmath25 as @xmath26 where @xmath27 is a random angle distributed uniformly over a range of angle @xmath28 .
this range describes qualitatively the random character of the collisions and @xmath29 can also be called the surface micro - roughness . in the following ,
the @xmath29 values are expressed in units of degree .
a few non - specular trajectories computed using the above rule , are shown in fig.[demo1]@xmath30 . in the case of specular reflections
, the reflection angle is not determined from numerical computation of trigonometric functions since the elements of the pre - fractal geometry are just parallel to the @xmath22 and @xmath31 axes . consequently , there is no numerical error due to truncation of angle computations .
the only truncation errors come from the solution of the linear equations . because of this error , there could exist trajectories , besides real ray splitting between close trajectories hitting salient corners , which are modified by a `` spurious '' splitting due to number truncation . to avoid spurious beam splitting , the cumulative error on the point of collision due to the numerical precision of the computer is calculated . for specular reflections ,
the initial positions of the particles are chosen in such a way that this error is always less than the distance between the point of collision and the nearest corner . for non - specular reflections ,
those trajectories are rejected and the particles are resent from the same position with a new random number sequence .
it should be emphasized that for @xmath17 , the number of specular reflections for the particular koch curve considered here can be determined from a simple recursion relation without solving the linear equations for the trajectory and the surface elements .
the number of collisions up to the third generations of the pre - fractal surface are listed in table @xmath32 . in the first generation
, all the particles send from the central part of the slit have @xmath33 collisions whereas the particles from the other two parts have only @xmath32 collision each .
the collision numbers are then in the form of @xmath34,@xmath33,@xmath35 in the first generation . in the next pre - fractal generation , each part of the the slit is again subdivided into another @xmath33 parts .
it can be seen that for @xmath36 , the collision numbers @xmath37,@xmath38,@xmath39 for the particles launched from the central part of the slit is just @xmath40,@xmath33,@xmath35 where @xmath33 is the number of collisions corresponding to the same part of the slit for the previous generation .
the other two sequences @xmath34,@xmath33,@xmath35 for the two other parts of the slit are given by @xmath41,@xmath33,@xmath35 .
the trajectories from these two parts had a single collision each at the previous generation .
the multiplication of the collision numbers from the first to the second generation of the pre - fractal is shown in fig.[demoe ] .
thus , the number of collisions at any two consecutive pre - fractal generations are given as : @xmath42 a recursion relation for the collision numbers for @xmath43 generation then can be written as @xmath44 where @xmath45 is the number of collisions of the trajectory from the same part of the slit at the @xmath18th generation . using the above recursion relation ,
the exact specular collision numbers for @xmath19 trajectories at any generation @xmath18 can then be obtained for @xmath17 .
the collision numbers and their spatial distribution along the slit are computed for the @xmath46th and @xmath47th pre - fractal generation ( @xmath48 and @xmath47 ) . a number of @xmath49 equally spaced particles are launched from the slit .
the spatial distribution of the collision numbers is studied under two different conditions .
firstly , the distribution of the specular collision numbers at @xmath50 angle of incidence for which the exact collision numbers are available , is investigated .
secondly , the effect of the surface micro - roughness @xmath51 on the collision spectrum is studied for the same slit angle . in this case , the collision spectrums are averaged over @xmath52 samples .
the general quality of the numerical results is verified with the exact collision spectrum for @xmath17 and @xmath53 obtained from the recursion relation ( eq .
[ seq ] ) . before discussing the multifractality associated with the spatial distribution of the collision numbers ,
it is important to consider their statistical distributions .
the probability to have a trajectory with @xmath54 collisions is measured by @xmath55 where @xmath56 is the number of trajectories having @xmath54 collisions and @xmath57 is the total number of trajectories .
the distribution of @xmath58 for different @xmath29 in fig .
[ pdist ] .
it can be observed that most of the trajectories exhibit between @xmath33 and @xmath38 collisions .
a large number of trajectories then leave the pre - fractal surface making only a few collisions and only a few trajectories make a large number of collisions .
it is important to note that the roughness @xmath29 has little effect on the collision number distribution except the reduction in the maximum number of collisions with increasing @xmath29 .
the total number of collisions is also very weakly dependent on the randomness @xmath29 .
these features have already been observed in the case of interaction of ballistic trajectories with pre - fractal pores @xcite in the knudsen diffusion regime where the collision number distribution was found to be of the lvy type .
however , up to the @xmath47th pre - fractal generation , no simple power law distribution of the collision numbers was found . to study the multifractal aspect of the trajectory distribution
, a collision spectrum is obtained .
the collision spectrum is a distribution of the relative collision numbers of the trajectories over the slit .
the relative collision number @xmath1 of a particle trajectory originating from a position @xmath22 on the slit and making @xmath21 collisions with the surface is given by @xmath59 where the sum is over all the particles .
this normalized collision number is the measure under study here . in fig .
[ mux1 ] , @xmath1 is plotted against the initial particle position keeping @xmath60 ( specular reflection ) for @xmath48 .
the value of @xmath1 is represented by impulses at the initial positions @xmath22 .
the spectrum is symmetrical around the central position , as expected .
most interesting to note , this spectrum corresponds to a uniform exploration of the irregular surface as there is a single collision per segment ( see fig .
[ demoe ] ) . the maximum number of collisions made by the central trajectory is @xmath61 and the total number of collisions made by all the particles is @xmath62 ( the first generation has @xmath63 collisions ) at @xmath48 .
thus , @xmath64 is given by @xmath65 , as found numerically .
the collision spectrums obtained in the presence of surface micro - roughness , @xmath66 ( @xmath67 ) , @xmath68 ( @xmath69 ) , and @xmath70 ( @xmath71 ) are shown in fig .
[ mux2 ] for the @xmath48th pre - fractal generation . due to micro - roughness ,
the spectrums become more uniform and the value of @xmath64 is decreased with respect to the specular value .
the value of @xmath64 is decreasing with increasing @xmath29 as it is also seen in the collision number distribution ( fig .
[ pdist ] ) that the maximum number of collisions is decreasing with @xmath29 . to show the multifractal nature of the distribution @xmath1 , it is necessary to study the scaling of the @xmath0-moments of the measure over different length scales on the slit .
if the measure @xmath1 is multifractal and the slit is divided into @xmath72 boxes of size @xmath73 , then the weighted number of boxes @xmath74,@xmath75 is given by @xmath76 where @xmath77 is the sum of the relative collision number of trajectories initiated in the @xmath78th box . here
@xmath2 is called the `` collision exponent '' . in fig .
[ expo ] , the values of @xmath2 , obtained by the box counting method , are plotted against @xmath0 for @xmath48 ( circle ) and @xmath47 ( square ) with @xmath60 ( specular reflections ) .
the values of @xmath2 do not depend on @xmath18 , the macro - irregularity of the surface .
note that , @xmath79 is @xmath80 and @xmath81 is @xmath82 here .
@xmath79 corresponds to the dimension of the slit which is @xmath32 here and @xmath81 is zero because @xmath83 .
the fact that a sequence of exponents are obtained irrespective of the generation @xmath18 confirms the multifractal nature of the spectrum .
note that the multifractality obtained for the specular collision spectrum at @xmath50 angle of incidence corresponds to a uniform exploration of the irregular surface ( fig .
[ demoe ] ) .
consequently , a measure defined over the irregular surface instead of the slit will not exhibit multifractality .
a check of the validity of our numerical computations can be made for the case @xmath17 with specular reflections . in that case there
exists a trinomial multiplicative cascade @xcite .
a comparison between the collision spectrum and the multiplicative cascade is made here . at the zeroth generation of the cascade @xmath84 ,
a uniform distribution of unit mass over an interval @xmath85,@xmath86 $ ] is considered . in the next generation @xmath87 ,
the unit mass is distributed over three equal intervals @xmath88 $ ] , @xmath89 $ ] , and @xmath90 $ ] with weighted probabilities @xmath91 , @xmath92 , and @xmath93 similar to the collision number distribution over the slit at @xmath17 . at @xmath94 generation
, the mass distribution will be exactly the same as the collision number distribution over the slit obtained for @xmath48 generation of the pre - fractal surface .
the measure over the unit interval remains always conserved in the iteration process since @xmath95 .
the mass exponent @xmath96 can be calculated in terms of @xmath97 as @xmath98^k}{\ln3^{-k } } = - \frac{\ln[p_0^q+p_1^q+p_2^q]}{\ln3}.\ ] ] the solid line in fig .
[ expo ] represents the mass exponent @xmath96 for this trinomial mass distribution . with no surprise
, there exists an excellent agreement between the exponents for the specular collision number distribution at @xmath17 and the trinomial mass distribution .
the effect of micro - roughness ( @xmath29 ) on the collision exponents is discussed now . in fig .
[ tqrg ] , @xmath2 is plotted against @xmath0 for @xmath66 ( circle ) , @xmath99 ( square ) , and @xmath100 ( triangle ) . the solid line represents the collision exponents for @xmath53 , the specular reflections . at large positive @xmath0
, the micro - roughness @xmath29 has a significant effect on @xmath2 with respect to @xmath53 .
it is expected because the spectra change with @xmath29 . again a sequence of collision exponents @xmath2 is obtained whatever the micro - roughness @xmath29 .
this means that the spectrum are still of multifractal nature . to determine the nature of the multifractality , the dimensions @xmath3 are obtained as usual as a function of the lipschitz - h@xmath101lder exponent @xmath102 through a legendre transformation @xcite @xmath103 of the sequence @xmath2
the curve @xmath3 for specular reflexions is discussed first . in fig .
[ fa45 ] , it is represented by circles .
the solid line represents @xmath3 for the trinomial mass distribution .
naturally , there is a good agreement between @xmath3 of specular collision distribution and the trinomial mass distribution .
since the slit here is a one dimensional object , the @xmath3 curves are always @xmath104 and @xmath105 is equal to @xmath32 .
one observes two parabolas in the @xmath3 curve .
this means that the multifractal is of trinomial nature .
the values of @xmath4 and @xmath5 for specular reflections at @xmath17 can be predicted .
the increment in the measure over a length @xmath106 to @xmath107 is @xmath108 .
the lipschitz - h@xmath101lder exponent @xmath109 is defined as @xmath110 .
after @xmath111th iteration , @xmath112 and @xmath109 is given by @xmath113 where @xmath114 and @xmath115 for the case considered here .
@xmath4 and @xmath5 correspond to @xmath116 and @xmath117 respectively and they are obtained as @xmath118 and @xmath119 .
these two extreme values are marked by two crosses on the @xmath102-axis .
it can be seen that @xmath4 and @xmath5 obtained numerically for the specular collision spectrum at @xmath17 are very close to the predicted values . note that , as expected , the whole @xmath3 versus @xmath102 is independent of the generation @xmath18 of the pre - fractal surface .
this is shown in fig .
[ fa45 ] , where the data for @xmath48 ( circles ) and @xmath120 ( squares ) collapse onto the same curve .
it is also interesting to notice that @xmath5 is exactly equal to the dimension @xmath10 of the pre - fractal and @xmath121 .
these features are specific of the prefractal geometry and the angle of incidence considered here . always in the case of specular reflections , but
for other angles of incidence like @xmath122 and @xmath123 , it is found that the @xmath3 versus @xmath102 curves also consist of two parabolas .
however , in both cases , the values of @xmath4 and @xmath5 keep on changing with the generation @xmath18 upto @xmath47th generation .
this is attributed to the fact that the multiplicative process in these cases is more involved than in the case of @xmath50 incidence .
there should exist for these cases a cascade of different sequential values of @xmath124 , @xmath125 and @xmath126 so that the asymptotic multifractal measure is still not reached for @xmath127 .
the effect of micro - roughness of the surface on the @xmath3 curve is discussed now . in fig .
[ fadf ] , @xmath3 versus @xmath102 curves are plotted for @xmath66 ( circles ) , @xmath99 ( squares ) , and @xmath100 ( diamonds ) at @xmath17 and @xmath48 .
the solid line represents the @xmath3 curve for specular collisions .
the triangles represent the results for an hypothetic `` totally '' diffusing surface for which re - emission would be isotropic , @xmath8 ; occuring randomly in any direction between @xmath128 to @xmath129 with respect to the normal , independent of the angle of incidence .
this is equivalent to the @xmath130 limit .
there are two important things to note .
first , the micro - roughness ( or random character of the particle re - emission ) of the surface is unable to destroy either the multifractality or the trinomial nature of the multifractality even at the maximum possible value of @xmath29 , corresponding to the `` totally diffusing surface '' .
second , the values of @xmath4 depends strongly on the degree of surface roughness @xmath29 whereas @xmath5 remains almost constant .
the value of @xmath4 increases as a function of @xmath29 .
this is expected . as @xmath29 increases , the value of @xmath64 decreases , the spectrum becoming more uniform as seen in fig .
[ mux2 ] and the smaller value of @xmath64 corresponds to the larger value of @xmath4 .
the value of @xmath5 could also be estimated from the knowledge of the minimum measure @xmath131 . for a single realization ,
the total number of collisions is @xmath132 .
thus , @xmath133 ( for a trajectory having a single collision ) and the value of @xmath5 is then given by @xmath134 very close to what is obtained here .
since the value of @xmath131 is almost independent of the surface roughness ( total number of collisions almost independent of @xmath29 ) in this collision process , the value of @xmath5 remains unchanged . from these results
, it is found that the multifractality as well as the trinomial nature of the multifractality associated with this collision process remains unaltered irrespective of the surface micro - roughness .
this is because , the multifractality of the collision spectrum is the outcome of a multiplicative cascade of the spatial sub - distributions which are independent of the various situations considered here .
the trinomial nature of the multifractal is due to the type of pre - fractal chosen here to model the surface macro - irregularity : a quadratic koch curve , each element being divided into three elements in the next generation .
a different pre - fractal generator , @xmath8 ; a different model of surface macro - irregularity will then lead to a different multifractal behavior .
the information dimension @xcite is the value of @xmath3 when @xmath135 .
the information dimension @xmath6 is found to be @xmath136 for @xmath50 angle of incidence with @xmath60 and at @xmath48 .
this indicates that the measure is supported by a minor fraction of the particle trajectories .
this also illustrates the fact that most of the trajectories interacts weakly with the fractal surface .
interestingly , the information dimension increases with the random character of the particle - surface interaction . indeed for @xmath66 , @xmath99 and @xmath100 the values of information dimensions are @xmath137 , @xmath138 and @xmath139 respectively . for the diffused surface @xmath140 .
this corresponds to more and more uniform collision number distribution on the slit .
it is interesting to comment on the uniform exploration of the surface , each element receiving the same number of particles at least for an angle of incidence of 45 degrees .
this is analogous to the results obtained by andrade et al @xcite through a molecular dynamics study of knudsen diffusion in the same geometry . in this last case , it is the first collision that is uniformly distributed over the perfectly absorbing surface whereas here , the surface is perfectly reflecting .
the uniform exploration found in these both cases constitutes a qualitative evidence of the absence of screening in knudsen diffusion .
it can be considered as a remarkable fact that , seen from the source of particles , the collision numbers are strongly non - uniform , while seen from the surface the exploration is uniform .
this may have interesting consequences for catalyst deactivation which should also be uniformly distributed .
the collision number distribution and their spatial behaviour are obtained for @xmath7 ballistic particles interacting with a pre - fractal curve of dimension @xmath10 .
the trajectories considered here correspond to the knudsen diffusion regime .
the micro - roughness of the surface , or the random character of the particle - surface interaction , is included through a randomness parameter @xmath29 in the reflection angle .
it is found that most of the particles make only a few collisions whereas a few trajectories exhibit a large number of collisions .
a multifractal measure of the spectrum is defined in terms of the relative collision number over a linear slit from which the trajectories are launched .
a sequence of `` collision exponents '' @xmath2 , is determined by the box counting method and it is found that the sequence depends on the moment @xmath0 .
thus , the spatial distribution of the collision numbers is a multifractal spectrum .
the fractal dimensions @xmath3 depend on the random character of the particle re - emission @xmath51 .
irrespective of the values of @xmath29 , the plot of @xmath3 versus lipschitz - h@xmath101lder exponent @xmath102 consist of two parabolas .
the spectrum is then that of a multiplicative trinomial multifractal independent of surface micro - roughness .
the information dimension @xmath6 of the measure is found to be smaller than @xmath32 and increases with the random character of the particle surface interaction .
different type of surface macro - irregularity would lead to different multifractal behavior .
notably , the surface however works uniformly , indicating an absence of screening in the knusdsen diffusion regime .
j. m. thomas and w. j. thomas , _ principle and practice of heterogeneous catalysis _
( vch , weinheim , 1997 ) ; g. f. froment and k. b. bischoff , _ chemical reactor analysis and design _ ( john wiley and sons , new york , 1990 ) .
s. b. santra and b. sapoval , phys .
e * 57 * , 6888 ( 1998 ) ; s. b. santra , b. sapoval , and o. haeberl , in _ frctals in engineering : from theory to industrial applications _ , edited by j. lvy - vehel , e. lutton , and c. tricot ( springer , london , 1997 ) .
s. feng and l. golubovic , phys .
65 * , 1028 ( 1990 ) ; l. golubovic , s. feng , and f. zeng , phys . rev . lett . * 67 * , 2115 ( 1991 ) ; j. p. bouchaud , d. touati , and d. sornette , phys . rev . lett . * 68 * , 1787 ( 1992 ) .
.[table ] number of specular collisions made by ballistic trajectories incident at @xmath50 and interacting with a quadratic koch curve of dimension @xmath10 at different generations @xmath18 .
the number of collisions at a generation @xmath142 is given by @xmath143 where @xmath45 is the number of collisions in the previous generation for the same trajectory . [
cols="<,^,^,<,^,^,<,^,^,<,^,^,<,^,^,<,^,^,<,^,^,<,^,^,<,^,^,^ " , ] | ballistic particles interacting with irregular surfaces are representative of many physical problems in the knudsen diffusion regime . in this paper ,
the collisions of ballistic particles interacting with an irregular surface modeled by a quadratic koch curve , are studied numerically .
the @xmath0 moments of the source spatial distribution of collision numbers @xmath1 are characterized by a sequence of `` collision exponent '' @xmath2 .
the measure @xmath1 is found to be multifractal even when a random micro - roughness ( or random re - emission ) of the surface exists .
the dimensions @xmath3 , obtained by a legendre transformation from @xmath2 , consist of two parabolas corresponding to a trinomial multifractal .
this is demonstrated for a particular case by obtaining an exact @xmath3 for a multiplicative trinomial mass distribution .
the trinomial nature of the multifractality is related to the type of surface macro - irregularity considered here and is independent of the micro - roughness of the surface which however influence the values of @xmath4 and @xmath5 .
the information dimension @xmath6 increases significantly with the micro - roughness of the surface .
interestingly , in contrast with this point of view , the surface seems to work uniformly .
this correspond to an absence of screening effects in knudsen diffusion . . |
the cosmic infrared background ( cib or cirb ) contains a considerable fraction of the collective radiation emitted by stars through the cosmic time .
stars in the epoch of reionization ( eor ) have the bulk of their radiation redshifted into the near - infrared band ( @xmath40.7 - 5 @xmath5 m ) and the cib measured in this band is specifically named near - infrared background " ( nirb ) . as such
, the nirb offers a unique opportunity to study faint high-@xmath0 galaxies that remain largely undetected in deep galaxy surveys ( see e.g. @xcite ) .
this is particularly important as these objects are commonly believed to provide most of the ionizing power that drives cosmic reionization @xcite . besides
, the nirb might also help characterizing the stellar populations of the first cosmic systems @xcite .
most recent studies have converged on the prediction that on scales of @xmath1 the fluctuation level from normal star - forming galaxies at @xmath6 is @xmath7 nwm@xmath8sr@xmath9 at 3.6 @xmath5 m @xcite . however , extracting such signal from available data has been so far very challenging , as even when the deepest galaxy subtraction from nirb maps is applied , the remaining flux fluctuations still can not be ascribed to the known high-@xmath0 galaxy population . on small angular scales ,
most of the signal arises from unresolved low-@xmath0 galaxies ( see the first analysis by @xcite ) . on larger scales
the measured power ( see e.g. @xcite ) is @xmath10 times larger than the low-@xmath0 galaxies @xcite , and @xmath11 times larger than that expected from high-@xmath0 normal star - forming galaxies and first stars @xcite .
therefore it must be attributed to some , yet unknown , alternative sources .
basically , two different explanations have been proposed for the origin of such large scale ( @xmath12 ) power excess " .
they involve either early accreting black holes @xcite which could explain the detected nirb - cosmic x - ray background coherence @xcite , or `` intrahalo light '' that radiated by stars ejected from their parent galaxies during merger events @xcite . at present
it is unclear which hypothesis should be preferred . besides , it is also possible that first stars or black holes are much more abundant than in the standard theoretical framework for example if the slope of the density fluctuations power spectrum sightly deviates from the standard one , the number of small halos ( thus stars or black holes therein ) could be boosted exponentially they can then provide sufficient radiation power . to be consistent with electron scattering reionization bounds , at the same time
the ionizing photon escape probability must be rather low ( for a detailed discussion see @xcite ) . to further complicate the interpretation ,
recent observations @xcite show that , on large scales and at least for the 1.1 and 1.6 @xmath5 m bands , diffuse galactic light ( dgl ) might provide a non - negligible flux contribution
. however @xcite ( see @xcite as well ) found that at the 3.6 and 4.5 @xmath5 m bands the dgl or the galactic cirrus component is largely subdominant .
for the above reasons , it is urgent to devise new strategies that put our understanding on firmer grounds . in order to isolate the signal from increasingly high redshifts , with sufficient depth and angular resolution one can in principle remove foreground galaxies down to extremely faint levels .
this is challenging for current instrumentation , and also not an easy task for the jwst , as the signal from high-@xmath0 galaxies is expected to be subdominant even at @xmath432 ab mag @xcite .
alternatively , cross - correlation studies seem promising . the nirb - hi 21 cm line cross - correlation @xcite has the advantage that it selectively picks up the signal from reionization sources . also , the nirb , if produced by sources in the eor , would be cross - correlated with the cmb through the thermal sunyaev - zeldovich effect .
this correlation could be seen in the forthcoming euclid - based all - sky cib maps @xcite .
in addition to cross - correlation , a recent work @xcite proposed the use of lyman - break tomography to constrain the nirb contribution from sources above a certain redshift .
@xcite analyzed the cross - correlation between the acs - detected faint sources and the source - subtracted nirb .
they found significant correlation on small scales , implying that the faint acs galaxies do contribute to nirb fluctuations .
however , these sources ( or objects associated with them ) can not account for the large scale clustering as the measured correlation is negligible .
their analysis did not pay particular attention to the high-@xmath0 sources , therefore the information on high-@xmath0 galaxies can not be directly derived from there .
it is surprising that so far little attention has been devoted to the search for high-@xmath0 normal star - forming galaxy signatures in the nirb , given that deep galaxy surveys have made tremendous progresses in obtaining the uv luminosity functions ( lf ) up to @xmath13 @xcite , and the detection limits of lyman break galaxy ( lbg ) surveys carried out by hst have already reached @xmath14 @xcite . measuring their intensity fluctuations present ( though tiny ) in the nirb is essentially one of the most exciting perspectives as it might convincingly show that nirb fluctuation experiments can be used to study the first galaxies .
such an experiment is also complementary to more traditional galaxy surveys that derive the lf of the brightest galaxies among early populations . in summary , it is mandatory to show that the nirb power spectrum signal of already known high-@xmath0 galaxies can be recovered robustly .
demonstrating a successful strategy will represent a major step forward in the methodology and allows to obtain , in addition to the clustering signal , other key quantities , as e.g. the colors of high-@xmath0 galaxies beyond the observed @xmath15 band .
colors , in turn , provide potentially information on stellar ages and initial mass function of the stars harbored by galaxies in the eor .
the idea we propose here is to isolate the targeted lbg signal and show the feasibility of statistically detecting reionization sources by cross - correlating deep lbg surveys with nirb maps .
to this aim , we : ( a ) construct large scale mock maps of the source - subtracted nirb and lbg catalogs using semi - numerical simulations ; ( b ) perform a cross - correlation analysis between the two data sets to extract the contribution of high-@xmath0 galaxies in the nirb .
the paper is organized as follows . in sec .
[ method ] we describe the steps to construct the mock maps . in sec .
[ results ] we present the analysis about the correlation coefficient and the colors . conclusions and discussions are presented in sec .
[ conclusions ] . throughout this paper
we use planck cosmological parameters : @xmath16 and @xmath17 .
all magnitudes are in the ab - system @xcite .
using the code dexm @xcite we carry out semi - numerical simulations to get catalogs of halos with mass @xmath18 from @xmath19 to 10 for every @xmath20 . we adopt a 400 mpc box size that corresponds to an angular size of @xmath21 deg at @xmath22 .
we construct a cuboid by replicating the output boxes along the line - of - sight , adding random translations , rotations and reflections @xcite .
this is our light - cone since we assume that all lines - of - sight are parallel an assumption that is convenient and safe enough when @xmath6 . to construct flux maps from the light - cone
, we link galaxy luminosities to halo mass @xmath23 .
the observed lfs could be reproduced exactly if we derive the @xmath2 relation through abundance matching , i.e. we force the number density of galaxies with luminosity @xmath24 to match the number density of halos with mass @xmath25 .
formally , @xmath26 where @xmath27 is the uv lf at 1600 and we use the schechter parameterization with the redshift - dependent fitting parameters given in @xcite . as a reference , in the derived @xmath2 relations our minimum mass @xmath28 corresponds to an absolute magnitude @xmath29 at @xmath30 respectively .
luminosity at other uv wavelengths is obtained through the luminosity - dependent spectral energy distribution ( sed ) slope @xmath31 ( i.e. , @xmath32 ) in @xcite , the values of @xmath33 and @xmath34 at @xmath35 and 7 could be found in @xcite . for convenience of using this form at in - between redshifts , we fit @xmath0-dependent forms @xcite : @xmath36 and @xmath37 .
we use these fittings anyway when @xmath3 . ] .
however , generally speaking this power - law only holds at @xmath38 , while we need luminosities at least until @xmath39 m , say 7500 when @xmath40 .
therefore at @xmath412000 we use the sed template from starburst99 @xcite , adopting a continuous star formation mode , with metallicity 0.1 @xmath42 and 200 myr stellar age .
the sb99 seds are normalized to match the power - law form at 2000 .
the flux received in each pixel in the map is the sum of radiation from all galaxies seen by the pixel , @xmath43 where @xmath44 , @xmath45 is the redshift of the @xmath46-th halos is much larger than the uncertainties . ] in the solid angle @xmath47 ( we adopt @xmath48 for all mock maps in this work ) , @xmath49 is its comoving distance . in fig .
[ maps ] we show the 3.6 @xmath5 m flux map ( bottom left panel ) from all galaxies between @xmath50 ( flux from galaxies at @xmath51 is negligible compared with galaxies at lower redshift , we ignore it here ) .
finally , we construct the flux map of lbgs at @xmath52 . to take into account selection effects
, we assume a completeness function of the form @xmath53,\ ] ] where erf is the error function , and @xmath54 is the limiting magnitude .
when constructing the flux map , for each lbg with apparent magnitude @xmath55 , we generate an uniformly distributed number @xmath56 . flux from these galaxies is added to the map only if @xmath57 . in fig .
[ maps ] we show the 1.6 @xmath5 mm flux maps of lbgs , because we consider a redshift range from @xmath19 to 10 . for shorter wavelengths all procedures ( and conclusions )
are similar , with the only exception of a slightly smaller signal due to the lyman dropout of @xmath58 galaxies . ]
flux map constructed from the lbgs for @xmath15-band limiting magnitudes @xmath59 ( top left ) and @xmath60 ( top right ) respectively .
in addition to the flux from high-@xmath0 ( @xmath61 ) galaxies , the observed nirb also contains radiation from unresolved , low-@xmath0 galaxies , and an excess radiation from unknown sources @xcite .
we collectively refer to these two components as _ contamination _ , since in this work the targeted signal is the flux from high-@xmath0 galaxies .
to model such contaminating signal we construct random maps with mean flux 1.0 ( 0.7 ) nwm@xmath8sr@xmath9 at 3.6 ( 4.5 ) @xmath5 m and reproduce the sum of ( i ) the angular power spectrum of the power _ excess _ ( see @xcite ) matching available observations @xcite , and ( ii ) the angular power spectrum of low-@xmath0 galaxies @xcite producing shot noise level matching @xcite ( @xmath62nw@xmath63m@xmath64sr@xmath9 at 3.6 @xmath5 m and @xmath65nw@xmath63m@xmath64sr@xmath9 at 4.5 @xmath5 m , the corresponding subtraction magnitude is @xmath66 mag ) .
the steps are : * a white noise map , i.e. a gaussian random field , is generated .
* this map is then transformed into spatial - frequency space by fft . * for each complex number in frequency space , its modulus is rescaled to be @xmath67 , where @xmath68 is the given power spectrum and @xmath69 is the spatial - frequency .
the zero - frequency ( @xmath70 ) element is set to be the mean flux . *
the above map is then transformed back into real space by inverse fft , resulting in a synthetic image with the same 2-point clustering properties as the measured @xmath71 . in fig .
[ maps ] we plot a single realization of the contamination map at 3.6 @xmath5 m as an example ( bottom right ) .
the contamination is not correlated with the high-@xmath0 galaxy component ; however , it adds noise to the cross - correlation signal . to account for the statistical variance of the contamination ,
we make 30 independent realizations . in fig .
[ aps ] we show the angular power spectrum of bottom panels in fig .
all maps are convolved with a circular symmetric spitzer psf before further analysis .
we first analyze the correlation coefficient between the lbg flux map and the nirb map .
it is defined as @xmath72 where @xmath73 refers to either 3.6 @xmath5 m or 4.5 @xmath5 m , @xmath74 and @xmath75 are the flux fluctuations of the same pixel ( zero lag ) at those two wavelengths .
the brackets refer to the pixel - averaged fluctuations .
the correlation coefficient indicates the fraction of sources contributing to common signals . before calculating the correlation coefficient we smooth both the nirb maps and the lbg flux maps by a real space top - hat window function with diameter @xmath76
this is to suppress the instrumental noise since in the measured nirb maps @xcite the instrumental noise is negligible at @xmath77 .
the smoothing is not related to the lbg detection limits since it is performed on maps constructed from lbgs already present in catalogs . to mimic surveys with different areas , we cut out sub - maps with different areas from the full map .
we choose three areas : @xmath78 , @xmath79 and @xmath80 deg@xmath81 , representing a survey region similar to hudf / xdf , uds , and an hypothetical larger field , respectively . before calculating the correlation coefficient , in both maps we mask pixels containing galaxies brighter than mag @xmath66 at _ either _ 3.6 @xmath5 m or 4.5 @xmath5 m . from this procedure
we obtain the source - subtracted nirb map . the correlation coefficient vs. lbg limiting magnitude
is shown in fig .
[ correlation_coefficient ] .
the filled regions contain 68.3% probability of all sub - map samples .
note that we have 30 random realizations for each full map ( ( 2.4)@xmath81 deg@xmath81 ) , so even for the @xmath80 deg@xmath81 case there are 120 samples , sufficient to compute the correlation coefficient variance .
[ correlation_coefficient ] shows that it is indeed feasible to detect the cross - correlation signal from the mock maps , even for a relatively shallow survey with @xmath82 , for which @xmath83 . by pushing the limiting magnitude to fainter values ,
the correlation coefficient rapidly increases by a factor @xmath84 , and then slowly approaches @xmath85 at @xmath86 .
it is worth noting that in small fields the measured correlation coefficient has a @xmath87 relative field - by - field scatter for @xmath88 , and even larger scatter at @xmath89 . in some extreme cases
there would be no or negative cross - correlation detected , due to the small number of lbgs contained in the fields .
to elucidate the differential contribution of lbgs , for a @xmath79 deg@xmath81 field , we further show in fig .
[ correlation_coefficient_z ] the correlation coefficient from lbgs at @xmath90 for @xmath91 and 27 , respectively .
for example , lbgs at @xmath92 contribute @xmath93 at @xmath94 .
our results are for nirb maps with a fixed shot noise level matching @xcite measurements .
since the shot noise is mainly from low-@xmath0 galaxies , an increased subtraction depth implies a relatively larger number of low-@xmath0 galaxies ( with respect to the high-@xmath0 ones ) are resolved and removed . as a result
we expect a higher correlation coefficient . on the other hand ,
if we resolve more lbgs , the cross - power signal also becomes stronger .
thus , the strength of the cross - power signal vs. different lbgs detection depths can be used to determine the differential contribution of early galaxies .
so far we have investigated the behavior of the correlation coefficient by at a specific smoothing scale @xmath76 .
we now examine its dependence on angular scale . to this aim
we re - define the correlation coefficient in the spatial - frequency domain via the power spectrum @xmath95 where @xmath96 is the cross - power spectrum and @xmath97,@xmath98 are the auto - power spectra .
they are all calculated using the 2d fft . since @xmath99 , @xmath100 for the same @xmath101 .
first we derive the shot noise due to lbgs fainter than a given h - magnitude ( fig .
[ psn ] ) .
it is the value of @xmath102 taken on a sufficiently small scale @xmath103 ( a stability check of the results for scales up to @xmath104 has been performed ) . as we will point out in the following , shot
noise is the most important term contributing to the cross - correlation .
the shot noise level shown here is a useful reference for possible follow - up work on real data .
the @xmath105 between the source - subtracted nirb at @xmath106 m and the 1.6 @xmath5 m flux map of lbgs with @xmath107 and 27 is shown by fig .
[ correlation_coefficient_q ] .
the errorbars are the r.m.s of 30 samples each one using different random contamination realizations .
the cross - power spectrum includes both the shot noise and clustering terms .
shot noise is from sources common to both nirb and lbg flux , and it is dominant on scales @xmath108 . on larger scales ,
the clustering term progressively takes over .
this term arises from all sources sharing the same large scale structure , including galaxies even fainter than the lbg limiting magnitude . as a consequence , in principle
the clustering term could allow the detection of fainter galaxies in the source - subtracted nirb through the cross - correlation with relatively bright lbgs .
however , as shown by fig .
[ correlation_coefficient_q ] the cross - correlation is more easily detected on small scales . although the clustering term may contain more information , the main difficulty to be overcome in order to efficiently use this strategy is that even for a relatively large survey area of @xmath109 deg@xmath81 ( i.e. our full map ) , the signal - to - noise ratio never exceeds @xmath110 at @xmath111 for @xmath112 . while increasing the limiting magnitude to @xmath113 does not help much to this aim
, the noise could be reduced by using larger survey areas , as expected with , e.g. euclid and wfirst . from our mock maps
it is also possible to characterize the colors , i.e. the ratio between flux in two bands , of unresolved galaxies in nirb observations . with the assumption
@xmath114 we derive the magnitude difference as @xmath115-[4.5 ] & = 2.5{\rm log}\left ( \frac{4.5f_{4.5}}{3.6f_{3.6 } } \right ) \nonumber \\ & \approx2.5{\rm log}\left ( \frac{4.5}{3.6 } \frac{\mean { ( \delta f_{4.5 } \delta f_{1.6 } ) _ \theta } } { \mean { ( \delta f_{3.6 } \delta f_{1.6 } ) _ \theta } } \right ) \end{aligned}\ ] ] such magnitude difference , which might considerably vary for individual galaxies , represents the combined and weighted color of the unresolved galaxy population in these bands . in practice the psf adds a bias to the measured flux ratio . therefore before calculating the magnitude difference it would be necessary to first deconvolve the psf from each map .
we skip this step here and directly calculate the magnitude difference on maps without psf convolution .
again we specify @xmath76 . the predicted color as a function of @xmath116 is reported in fig .
[ color ] , allowing us to conclude that the magnitude difference which is @xmath4 - 0.13 mag , could be detected by cross - correlating survey maps with area @xmath117 deg@xmath81 .
the figure reiterates that smaller area fields would be affected by bias effects .
motivated by the fact that lbg surveys carried out by the hst have reached detection limits much deeper than the nirb measured by spitzer at longer wavelengths , in this paper we investigated the feasibility to pick the resolved lbgs component out of the nirb by a cross - correlation analysis .
our investigations were based on mock maps constructed from semi - numerical simulations of halo formation , and empirically determined @xmath2 relations .
we found that in the nirb observed at 3.6 and 4.5 @xmath5 m and with sources subtracted down to @xmath425 apparent magnitude , at smallest scales where the shot noise dominates , about 10% of the flux fluctuations arises from lbgs in @xmath3 and with @xmath118 .
such faint galaxies have already been resolved in the existing deep surveys .
however , this fractional contribution , if measured from narrow fields with area @xmath119
deg@xmath81 ( as the hudf / xdf ) , could vary from @xmath120 to @xmath121 . if the limiting magnitude is decreased to @xmath122 , the fractional contribution decreases to about 8% .
if in this case we consider a larger field , e.g. with area similar to the uds ( @xmath123 deg@xmath81 ) , then the correlation coefficient varies in the narrower range @xmath124 .
we remind that the variance at hand here is due to both the large - scale inhomogeneity of the signal and the contamination : it is the field - to - field variance of the correlation itself .
we do not model errors introduced by data analysis procedures , as for example mask effects .
however , at least theoretically we have shown that the contribution from the faintest galaxies could be isolated from the nirb through the cross - correlation analysis .
we pointed out that it is still challenging to use the cross - correlation arising from the clustering term of the cross - power spectrum to study galaxies unresolved not only in nirb observations , but also in lbg surveys .
this term is dominant at @xmath125 , but even for survey areas as large as @xmath109 deg@xmath81 , the signal has a small significance , with a s / n ratio @xmath126 .
although very unlikely , if the nirb clustering excess originates from lbgs we would detect a correlation coefficient close to 1 at scales where the clustering term dominates .
stated differently , a measured correlation coefficient @xmath127 would actually rule out this scenario .
our results have interesting implications also for the colors of high-@xmath0 galaxy population .
the colors of lbgs detected at wavelengths @xmath128 m but too faint to be individually resolved by existing telescopes at 3.6 and 4.5 @xmath5 m , can be obtained by using the cross - correlation with the nirb in these two longer wavelength bands .
it is worth noting that the predictions presented in this paper assume a contaminating signal in the form of nirb fluctuation excess which originates from direct collapse black holes at high @xmath0 @xcite .
if however the excess is found to arise more locally , we might be able to model or subtract it more accurately , thereby making the signal calculated in this paper more easily detectable , i.e. @xmath129 . importantly , the study proposed here can be further developed to infer the properties of even fainter high-@xmath0 galaxy populations that are currently inaccessible to direct telescopic detection .
this could , for example , be accomplished by a lyman - break tomography study designed to isolate high-@xmath0 populations via multi - band cross - correlations without requiring prior lbg detections @xcite .
we thank a. kashlinsky for valuable comments on the manuscript . | emission from high-@xmath0 galaxies must unquestionably contribute to the near - infrared background ( nirb ) . however , this contribution has so far proven difficult to isolate even after subtracting the resolved galaxies to deep levels .
remaining nirb fluctuations are dominated by unresolved low-@xmath0 galaxies on small angular scales , and by an unidentified component with unclear origin on large scales ( @xmath1 ) . in this paper , by analyzing mock maps generated from semi - numerical simulations and empirically determined @xmath2 relations , we find that fluctuations associated with galaxies at @xmath3 amount to several percent of the unresolved nirb flux fluctuations .
we investigate the properties of this component for different survey areas and limiting magnitudes . in all cases ,
we show that this signal can be efficiently , and most easily at small angular scales , isolated by cross - correlating the source - subtracted nirb with lyman break galaxies ( lbgs ) detected in the same field by hst surveys .
this result provides a fresh insight into the properties of reionization sources .
cosmology : diffuse radiation - dark ages ; reionization , first stars infrared : diffuse - background galaxies : high - redshift |
fixons @xmath1 et @xmath2 deux entiers et @xmath3 .
soient @xmath4 et @xmath5 deux systmes de variables . on se donne @xmath6 .
notons @xmath7 lanneau des oprateurs diffrentiels coefficients dans @xmath8 .
pour @xmath9={\mathbb{c}}[s_1,\ldots , s_p]$ ] , considrons lidentit suivante : @xmath10 f^{s+v},\ ] ] o @xmath11 .
un polynme @xmath12 vrifiant une telle identit est appel polynme de bernstein - sato ( associ @xmath13 ) .
lensemble de ces polynmes forment un idal appel idal de bernstein - sato ( associ @xmath14 ) et quon note @xmath15 .
+ rappelons que cest i.n .
bernstein @xcite qui , dans le cas @xmath16 et o @xmath14 est polynomiale , a montr que lidal @xmath15 est non nul ( dans ce cas , il faut , dans @xmath17 , remplacer @xmath18 par lalgbre de weyl @xmath19 , i.e. lalgbre des oprateurs diffrentiels coefficients polynomiaux ) .
dans le cas o @xmath14 est analytique et toujours pour @xmath16 , la non nullit de @xmath15 revient j.e .
bjrk @xcite avec des mthodes similaires celles employes dans @xcite .
dans ce mme cas , citons m. kashiwara @xcite qui publia une autre preuve et dmontra en plus que le gnrateur unitaire de lidal de bernstein - sato est racines rationnelles .
pour @xmath20 , la preuve dans le cas polynomial est une gnralisation facile de celle de i.n .
bernstein , que lon peut trouver dans @xcite .
dans le cas analytique avec @xmath20 , la non nullit de @xmath15 a t dmontre par c. sabbah ( @xcite et @xcite ) .
citons la contribution de a. gyoja @xcite qui a repris la preuve de c. sabbah en montrant de plus que @xmath15 contient un lment rationnel non nul .
lobjet du prsent article est une mise au point de la preuve donne par c. sabbah . plus
prcisment , on peut dcomposer la preuve de c. sabbah en deux grandes tapes : la premire utilise des arguments similaires ceux employs par m. kashiwara dans le cas @xmath16 , la deuxime consiste essentiellement en un rsultat de finitude qui permet par la suite de se ramener au rsultat de la premire tape .
la seconde tape de la preuve de c. sabbah sappuie sur un ventail dit adapt ( @xcite prop .
2.2.1 et th .
a.1.1 ) do nt lexistence mriterait une mise au point technique supplmentaire ( voir le commentaire qui suit le th .
aussi , nous proposons dans cet article un nonc et une dmonstation plus lmentaires et plus constructifs de la seconde tape , qui vitent la notion dlicate dventail adapt .
afin de motiver les rsultats du prsent article , nous rappelons , sans entrer dans tous les dtails , la preuve de c. sabbah .
dans ce rappel , nous mettrons en vidence le rsultat de c. sabbah pour lequel nous allons donner un nonc et une preuve plus constructifs .
signalons que la plupart des notions ou notations introduites ci - dessous qui nous seront utiles dans la suite seront dtailles dans les sections suivantes . on note
@xmath21 lanneau des oprateurs diffrentiels coefficients dans @xmath22 . en suivant la mthode de b. malgrange @xcite , on fait agir
@xmath21 sur @xmath23 f^s$ ] . on note
@xmath24 lidal ( gauche ) annulateur de @xmath25 dans @xmath21 et @xmath26 le quotient @xmath27 .
pour chaque @xmath28 , notons @xmath29 la @xmath30-filtration de kashiwara - malgrange associe la variable @xmath31 et notons @xmath32 la ( multi)filtration de @xmath21 indexe par @xmath33 : pour @xmath34 , @xmath35 elle induit une filtration @xmath36 sur @xmath26 o pour tout @xmath34 , @xmath37 est limage de @xmath38 par la projection @xmath39 .
pour @xmath28 , identifions la filtration @xmath40 avec la forme linaire sur @xmath41 donne par @xmath42 ( o @xmath43 , @xmath44 et @xmath45 correspondent respectivement @xmath46 ) .
notons @xmath47 .
on identifie @xmath48 @xmath49 .
chaque @xmath50 de @xmath51 ( i.e. @xmath50 coefficients entiers ) donne lieu une filtration naturelle @xmath52 de @xmath21 et de @xmath26 indexe par @xmath53 donne par : @xmath54 o @xmath55 si @xmath56 .
+ + voici maintenant les deux tapes de la preuve donne par c. sabbah .
etape 1 : : ( @xcite th .
3.1.1 , voir aussi @xcite 2.9 et 2.10 ) pour tout @xmath57 , il existe un polynme @xmath58 $ ] dune variable tel que pour tout @xmath59 , on ait @xmath60 + la preuve de ce rsultat utilise des arguments analogues ceux employs par m. kashiwara @xcite dans le cas @xmath16 .
+ notons @xmath61 le polynme unitaire de plus bas degr satisfaisant lidentit prcdente .
la contribution de a. gyoja @xcite consiste dans le fait que @xmath61 est racines dans @xmath62 .
etape 2 : : nous allons introduire deux nouvelles filtrations sur @xmath26 .
+ 1 . soit @xmath63 un cne convexe rationnel dans @xmath64 . on note
@xmath65 lensemble des lments primitifs du @xmath66-squelette de @xmath63 ( i.e. @xmath67 si et seulement si la droite engendre par @xmath50 est dans le @xmath66-squelette de @xmath63 et les coefficients de @xmath50 sont des entiers sans facteurs communs ) .
pour tout @xmath68 , on note : @xmath69 ( voir la figure [ fig : vsigma ] o @xmath70 et @xmath63 est engendr par @xmath71 et @xmath72 : @xmath73 est reprsent par un oprateur @xmath74 do nt le diagramme de newton est dans le quadrillage . ) + , width=415 ] + pour tout @xmath34 , nous avons facilement les inclusions suivantes : @xmath75 la premire tant triviale et la seconde venant du fait que @xmath65 est inclus dans ladhrence de @xmath63 .
+ ( @xcite th .
a.1.1 et prop .
2.2.1 ) il existe un ventail @xmath76 ( dit ventail adapt @xmath36 ) constitu de cnes polydraux rationnels convexes tel que pour tout cne @xmath77 et pour tout @xmath34 , on ait : @xmath78 + cest ce thorme s1 do nt nous allons donner un nonc et une preuve plus lmentaires .
disons un mot de la preuve .
dans lappendice de @xcite en collaboration avec f.j .
castro jimnez , c. sabbah dmontre lexistence dun ventail @xmath76 dit adapt la filtration @xmath36 aprs quoi il montre dans la proposition 2.2.1 que pour tout cne dun tel ventail , on a lgalit prcdente .
la preuve de lexistence dun tel ventail mriterait une mise au point .
en effet , elle sappuie sur une division paramtre qui conduirait lapparition de sries formelles dans les variables @xmath79 de drivation . 2 .
soit @xmath80 la filtration indexe par @xmath33 dfinie par : @xmath81 comme consquence du thorme s1 , nous avons : 1 .
pour tout @xmath68 , @xmath82 o @xmath83 dsigne lensemble des lments primitifs du @xmath66-squelette de @xmath76 . 2 .
il existe @xmath84 tel que pour tout @xmath34 , on ait @xmath85 + faisons quelques commentaires sur la preuve de ce rsultat .
lassertion ( a ) de s2 dcoule trivialement de s1 , en effet : @xmath86 en ce qui concerne , lassertion ( b ) , c. sabbah dmontre que
si un ventail @xmath76 satisfait lnonc du thorme s1 alors la filtration @xmath80 est une bonne @xmath87 filtration ce qui , par un lemme usuel de comparaison entre bonnes filtrations fournit les inclusions voulues ( il faut noter que la premire inclusion de ( b ) est trivialement vraie et que cest la deuxime qui nous intresse ) .
+ finissons le rappel de la preuve .
notons @xmath88 la classe de @xmath66 dans le quotient @xmath27 .
posons @xmath89 par les assertions ( a ) et ( b ) de s2 , on constate que : @xmath90 ce qui signifie que @xmath91 .
nous allons maintenant noncer les rsultats principaux du prsent article .
on se donne un idal @xmath24 de @xmath92 ( anneau des oprateurs diffrentiels coefficients analytiques faisant intervenir @xmath93 variables ) .
dans @xcite , a. assi , f. castro jimnez et m. granger ont introduit lensemble @xmath94 des formes linaires @xmath50 pour lesquelles la filtration de @xmath92 naturellement associe est compatible avec la structure non commutative de @xmath92 ( cf .
paragraphe 1.2 ) .
ils ont tudi le comportement du gradu @xmath95 lorque @xmath50 varie dans @xmath94 .
considrons la relation sur @xmath94 telle que @xmath50 et @xmath96 sont en relation si les gradus @xmath97 et @xmath98 sont isomorphes ( @xmath99 dsigne lhomognis de @xmath24 do nt nous rappelons la dfinition plus loin )
. cette relation est une relation dquivalence donnant sur @xmath94 une partition constitue de cnes polydraux rationnels convexes .
leventail ainsi constitu est appel ventail de grbner ( analytique ) associ @xmath99 et not @xmath100 .
maintenenant , posons @xmath101 et reprenons les notations prcdentes .
comme nous le verrons , on peut naturellement inclure @xmath48 dans @xmath94 . on note alors
@xmath102 lventail obtenu comme restriction de @xmath100 @xmath48 .
voici les rsultats quon se propose de montrer ici .
[ theo ] pour tout cne @xmath63 de @xmath103 , @xmath78 lanalogue du corollaire s2 en dcoule pour ( a ) et ( b ) en remplaant le 1-squelette de @xmath76 par celui de @xmath103 .
en effet , la preuve du corollaire s2 fonctionne pour nimporte quel ventail ds que ce dernier satisfait le thorme s1 .
[ theo2 ] pour @xmath70 , il existe @xmath104 calculable partir de bases de grbner associes chaque cne de lventail @xmath103 ( ce calcul sera dtaill la section 3 ) tel que pour tout @xmath105 , @xmath106 pour @xmath107 , une gnralisation de ce rsultat semble se heurter des difficults techniques difficiles rsoudre . en rsum , voici lapport du prsent article : * un nonc et une preuve constructifs du thorme cl s1 ce qui fournit une approche plus lmentaire et plus constructive de ltape 2 de la preuve de c. sabbah et qui permet dviter la notion dventail adapt .
* pour @xmath70 , une preuve entirement constructive de ltape 2 dans la preuve de c. sabbah .
je signale que cet article constitue une partie de ma thse
@xcite dans laquelle on pourra trouver une autre preuve de lassertion ( a ) de s2 sans passer par le thorme [ theo ] .
+ + dcrivons le contenu de cet article .
+ dans une premire section , nous faisons un certain nombre de rappels concernant le thorme de division dans @xmath108 tel quil est nonc dans @xcite , les notions de base standard et dventail de grbner .
la section 2 est consacre la dmonstration du thorme [ theo ] .
dans la troisime section , nous dmontrons le thorme [ theo2 ] .
dans les paragraphes qui suivent , nous rappelons , sans donner les dmonstrations un certain nombre de notions et de rsultats qui nous seront utiles dans les sections 2 et 3 .
@xcite , nous introduisons lanneau @xmath108 .
cest avec cet anneau que les auteurs de @xcite ont introduit lventail de grbner analytique .
dans ce paragraphe et jusqu 1.3.2 , @xmath109 et @xmath92 dsigne lanneau des oprateurs diffrentiels coefficients dans @xmath8 . on dfinit
lanneau @xmath108 comme la @xmath8-algbre engendre par @xmath110 o les seules relations de commutation non triviales sont : @xmath111={\frac{\partial { c(x)}}{\partial x_{i } } } z \text { pour } i=1,\ldots , n \text { et } c(x ) \in { \mathbb{c}}\{x\}.\ ] ] dans lanneau @xmath92 , on note @xmath112 le degr total en les @xmath79 dun lment @xmath113 .
considrons la filtration associe .
elle stend naturellement @xmath108 en considrant le degr total en les @xmath114 et @xmath115 .
cette filtration fait de @xmath108 une algbre gradue : @xmath116 o @xmath117 , @xmath118 et @xmath119 .
+ on dit quun oprateur @xmath120 est homogne ( de degr @xmath121 ) si @xmath122 .
+ pour @xmath123 , on dfinit son homognis @xmath124 comme suit .
+ on crit @xmath125 et on pose @xmath126 o @xmath127 , ainsi @xmath128 est homogne de degr @xmath112 .
+ de plus pour un idal @xmath24 de @xmath92 , on dfinit @xmath99 comme lidal ( gauche ) de @xmath108 engendr par lensemble des @xmath128 avec @xmath113 parcourant @xmath24 .
nous rappelons les notions usuelles de filtrations sur @xmath92 et @xmath108 , le thorme de division dans @xmath108 de @xcite .
nous rappelons aussi les notions de bases standards et bases standards rduites minimales .
soit @xmath94 lensemble des formes linaires @xmath129 , @xmath130 avec pour tout @xmath131 , @xmath132 et @xmath133 .
on tend @xmath94 @xmath134 en posant @xmath135 .
+ pour @xmath120 ( resp .
@xmath123 ) , on crit @xmath136 ( avec @xmath137 pour @xmath138 si @xmath123 ) . on dfinit
le diagramme de newton de @xmath113 , not @xmath139 ( resp .
@xmath140 ) , comme lensemble des @xmath141 pour lesquels @xmath142 .
+ etant donn @xmath143 et @xmath144 ( ou @xmath123 ) , on dfinit lordre de @xmath113 par rapport @xmath50 comme @xmath145 .
cette ordre induit une filtration @xmath146 sur @xmath108 ou @xmath92 indexe par @xmath147 donne par : @xmath148 ainsi quun gradu associ @xmath149 .
+ pour @xmath120 , on note @xmath150 le symbole principal de @xmath113 , i.e. la classe de @xmath113 dans quotient @xmath151 avec @xmath152 .
si @xmath153 est un idal de @xmath108 , on a une filtration @xmath154 induite qui conduit un idal gradu de @xmath155 quon note @xmath156 et qui est engendr par lensemble des @xmath157 pour @xmath158 .
+ + pour une forme @xmath143 , on dfinit deux ordres : @xmath159 sur @xmath160 et @xmath161 sur @xmath162 : @xmath163 o @xmath164 est un bon ordre compatible avec laddition fix pour toute la suite . @xmath165 pour @xmath120 , on dfinit lexposant privilgi @xmath166 et le monme privilgi @xmath167 .
de mme pour @xmath168 et @xmath159 la place de @xmath161 .
notons que @xmath169 ds que @xmath170 est compatible avec laddition ce qui est le cas pour @xmath159 et @xmath161 .
rappelons le thorme de division dans @xmath108 de @xcite .
soit @xmath143 .
soient @xmath171 une famille dlments de @xmath108 .
notons @xmath172 la partition de @xmath162 dfinie partir des @xmath173 : * @xmath174 * @xmath175 pour @xmath176 * @xmath177 ( @xcite theorem 7)[theo : div ] pour tout @xmath120 il existe un unique @xmath178 tel que : 1 .
@xmath179 2 .
pour tout @xmath180 , si @xmath181 alors @xmath182 3 .
si @xmath183 alors @xmath184 on appelle @xmath185 le reste de la division de @xmath113 par les @xmath186 relativement @xmath161 . * @xmath187 * @xmath188 soit @xmath153 un idal de @xmath108 et @xmath189 . on dit que @xmath171 forment une @xmath161-base standard
de @xmath153 si pour tout @xmath113 dans @xmath153 , le reste de la division de @xmath113 par @xmath171 est nul .
considrons lensemble des exposants de @xmath153 , @xmath190 .
etant donns @xmath189 , les deux assertions suivantes sont quivalentes ( grce aux thorme de division ) : * les @xmath186 forment une @xmath161-base standard de @xmath153 * @xmath191 lexistence dune base standard est assure par le lemme de dickson qui dit que si une partie @xmath192 de @xmath193 ( @xmath194 ) vrifie @xmath195 ( ce qui est le cas de @xmath196 ) alors il existe @xmath197 fini tel que @xmath198 .
il est facile de voir quune base standard nest pas unique , cest pour cela quil existe une notion de base standard rduite minimale : soit @xmath171 une @xmath161-base standard de @xmath199 et soit @xmath200 pour @xmath180 . * on dit
quelle est minimale si pour toute partie finie @xmath201 de @xmath162 on a limplication suivante : @xmath202 * on dit quelle est rduite si les @xmath186 sont unitaires ( i.e. le coefficient du monne privilgi est @xmath66 ) et si pour tout @xmath203 : @xmath204 il existe une unique @xmath161-base standard rduite minimale dun idal @xmath199 .
notons que si lidal @xmath153 est homogne alors la base standard rduite minimale est constitue dlments homognes .
nous terminons ce paragraphe par un lemme qui illustre lintrt des bases standards rduites minimales .
nous lappliquerons dans la section suivante .
[ lem : utile ] soient @xmath205 et @xmath206 deux ordres sur @xmath162 qui permettent de faire des divisions dans @xmath108 ( par exemple @xmath207 avec @xmath71 et @xmath72 deux formes ) .
soit @xmath208 la base standard rduite minimale dun idal @xmath209 par rapport lordre @xmath205 .
supposons que pour tout @xmath203 , @xmath210 et @xmath211 soient gaux .
alors @xmath208 est aussi la base standard rduite minimale de @xmath153 par rapport @xmath206 .
nous omettons la preuve .
disons simplement que ce qui importe lors dune division , cest lensemble des @xmath212 et pas lordre @xmath213 .
ainsi , diviser par les @xmath186 par rapport @xmath205 ou @xmath206 donnera les mmes quotients et le mme reste .
dans ce paragraphe , nous rappelons le rsultat principal de @xcite dcrivant lventail de grbner dun idal .
nous introduisons aussi le @xmath30-ventail de grbner @xmath214 , ce sera lobjet principal des sections @xmath215 et @xmath216 .
[ [ section ] ] on se donne un idal @xmath24 de @xmath92 et on considre @xmath217 son homognis .
pour @xmath50 et @xmath96 dans @xmath94 , on dfinit la relation : @xmath218 cette relation est une relation dquivalence sur @xmath94 .
[ theo : event_grob]@xcite la partition sur @xmath94 donne par la relation dquivalence prcdente est constitue de cnes polydraux rationnels convexes .
cette partition quon note @xmath100 est appele ventail de grbner ( analytique ) de @xmath99 . + de plus pour chaque cne @xmath219 , il existe @xmath220 homognes tels que : * pour tout @xmath221 , @xmath222 et @xmath223 pour tout @xmath203 .
* pour tout @xmath224 , lensemble @xmath208 est la base standard rduite minimale de @xmath99 par rapport @xmath161 .
remarquons que la deuxime condition nous montre que sur un cne @xmath63 , lensemble @xmath225 des exposants de @xmath99 par rapport @xmath161 est constant lorsque @xmath50 parcourt @xmath63 .
+ [ [ section-1 ] ] a partir de maintenant , on travaillera dans @xmath21 et @xmath226 ( i.e. @xmath101 ) .
pour @xmath28 , on notera @xmath227 la forme linaire donne par : @xmath228 o @xmath43 et @xmath229 .
cette forme linaire donne lieu une filtration quon note aussi @xmath40 et qui nest rien dautre que la @xmath30-filtration de kashiwara - malgrange associe la variable @xmath31 ( rappelons que dans @xmath21 , les variables sont @xmath4 et @xmath230 ) . on note alors @xmath30 la multifiltration @xmath32 .
+ nous noterons @xmath231 le sous - ensemble des formes linaires @xmath50 scrivant : @xmath232 avec @xmath233 , aussi nous identifierons @xmath48 et @xmath49 . a partir de maintenant , pour @xmath234 , on notera @xmath52 la filtration associe @xmath50 ( conformment aux notations de @xcite ) .
notons que pour @xmath235 , @xmath236 est isomorphe un sous - anneau de @xmath226 , ainsi nous considrerons que tous les calculs se feront dans @xmath226 .
maintenant , considrons la restriction de @xmath100 @xmath48 .
ceci donne lieu au @xmath30-ventail de grbner de @xmath99 que lon note @xmath214 .
comme consquence , on obtient un analogue du thorme [ theo : event_grob ] : [ cor : v_event_grob ] pour chaque cne @xmath63 de @xmath214 , il existe @xmath220 homognes tels que : * pour toute @xmath221 , @xmath222 et @xmath223 pour tout @xmath203 . * pour tout @xmath224 , lensemble @xmath208 est la base standard rduite minimale de @xmath99 par rapport @xmath161
. on voit que pour chaque cne @xmath63 , il existe des @xmath186 qui , pour tout @xmath50 de @xmath63 , forment la base standard rduite minimale de @xmath99 par rapport @xmath161 .
sil ny pas de confusion possible , on lappelle * la base standard de @xmath99 associe @xmath63*. terminons ce paragraphe par quelques remarques et notations . un cne @xmath63 de @xmath214 nest pas ncessairement ouvert ( il peut tre `` semi - ouvert '' ) .
voici comment on dfinit @xmath65 : + on considre dabord ladhrence @xmath237 de @xmath63 .
il existe @xmath238 quon suppose primitifs ( avec @xmath239 pouvant tre plus grand que @xmath240 ) tels que : @xmath241 supposons @xmath242 minimal alors @xmath65 est lensemble @xmath243 .
lensemble @xmath244 nest alors rien dautre que la runion des @xmath65 avec @xmath245 ( remarquons quil est constitu dlments entiers car tout cne @xmath63 est rationnel ) .
on dfinit lintrieur de @xmath63 comme lensemble des combinaisons coefficients strictement positifs des @xmath246 . on note @xmath247 le
cne ouvert engendr par les @xmath246 .
dautre part si @xmath71 et @xmath72 sont dans @xmath48 , on note @xmath248 le cne semi - ouvert contenant @xmath72 mais non @xmath71 .
dans la suite , on crira @xmath103 la place de @xmath214 .
soit @xmath63 un cne de @xmath103 .
nous avons vu dans le corollaire [ cor : v_event_grob ] quil existe une famille @xmath171 de @xmath99 qui est la base standard rduite minimale de @xmath99 par rapport @xmath161 et ce pour tout @xmath50 dans @xmath63 .
maintenant , que se passe - t - il pour une forme linaire @xmath50 appartenant ladhrence de @xmath63 mais qui nest pas dans @xmath63 ( ce qui peut arriver pour un lment de @xmath249 ) ?
bien entendu les @xmath186 ne forment pas ncessairement une base standard de @xmath99 pour lordre @xmath161 . par contre il est possible , et cest le but du rsultat suivant , de construire un ordre que nous noterons @xmath250 ( car il dpend de @xmath50 et de @xmath63 ) pour lequel les @xmath186 forment la base standard rduite minimale de @xmath99 .
voici comment on dfinit lordre en question .
+ dabord on fixe une forme linaire @xmath251 dans lintrieur de @xmath63 et pour @xmath252 et @xmath253 dans @xmath254 on pose : @xmath255 @xmath256 soit @xmath235 appartenant ladhrence de @xmath257 et soit @xmath171 la base standard de @xmath99 associe au cne @xmath63 .
alors pour tout @xmath180 et pour tout @xmath96 dans @xmath258 : @xmath259 comme * consquence * de cette proposition , nous obtenons que * les @xmath186 forment la base standard rduite minimale de @xmath99 pour lordre @xmath250*. en effet , il suffit pour cela dappliquer le lemme [ lem : utile ] .
en dautres termes , il est possible de faire baisser lordre par rapport lun des @xmath246 sans augmenter lordre par rapport aux autres @xmath246 .
avec ce lemme , nous sommes en mesure de donner une soit @xmath275 alors pour @xmath276 , il existe @xmath277 tel que @xmath278 et @xmath279 . on pose @xmath280 .
en appliquant un nombre fini de fois le lemme avec @xmath281 ( la premire fois avec @xmath282 et @xmath283 ) , on construit @xmath284 tel que @xmath285 et @xmath286 pour @xmath287 . on recommence le processus avec @xmath288 et @xmath289 , etc .
aprs un nombre
fini dtapes on obtient @xmath290 tel que @xmath291 et pour tout @xmath276 , @xmath292 .
ceci montre bien que @xmath293 .
par hypothse , il existe @xmath298 avec @xmath299 et @xmath300 .
il existe @xmath301 tels que @xmath302 . on pose alors
@xmath303 , @xmath304 et @xmath305 et comme @xmath306 , on a @xmath307 . + considrons la division de @xmath308 par les @xmath186 relativement lordre @xmath309 : @xmath310 o les @xmath311 forment la partition de @xmath312 associe aux exposants privilgis des @xmath186 ( voir le thorme de division [ theo : div ] rappel la section @xmath66 ) .
+ comme pour tout @xmath313 , les exposants @xmath314 et @xmath315 sont gaux , la division prcdente est aussi une division relativement aux ordres @xmath316 .
par consquent , pour tout @xmath276 et @xmath180 : @xmath317 notons @xmath153 lensemble des @xmath318 pour lesquels @xmath319 , nous avons alors : @xmath320 on considre et on note @xmath321 .
cest un lment de @xmath99 .
+ posons @xmath322 .
nous allons montrer les deux assertions suivantes . 1 . on a clairement @xmath326 .
par consquent , @xmath327 on voit alors facilement que les deux termes entre parenthses ont un @xmath71-ordre strictement infrieur celui de @xmath328 .
2 . fixons @xmath329 entre @xmath215 et @xmath242 .
+ en utilisant [ cor : v_event_grob ] , pour tout @xmath203 , on a : @xmath330 dautre part , par construction des @xmath331 , pour tout @xmath203 on a : @xmath332 , ce qui entraine ceci pour tout @xmath203 : @xmath333 maintenant , nous avons @xmath334 .
par la relation ( [ ident2 ] ) , on peut dire que cette criture est le rsultat de la division de @xmath335 par @xmath336 relativement @xmath337 , par consquent : @xmath338 de la mme manire , on peut montrer que @xmath339 par consquent , grce ( [ ident3 ] ) , ( [ ident4 ] ) et ( [ ident1 ] ) , on obtient lgalit @xmath340 , ce qui donne en particulier @xmath341 .
do les galits et ingalits suivantes : @xmath342 par consquent , @xmath324 .
les deux assertions sont dmontres .
maintenant , effectuons la spcialisation @xmath343 ( qui est un morphisme dalgbres @xmath344 ) et posons @xmath345 . on a @xmath346 donc @xmath347 et @xmath348 .
aprs spcialisation , les assertions @xmath66 et @xmath215 deviennent : @xmath273 pour tout @xmath276 avec ingalit stricte pour @xmath349 . le lemme est dmontr .
dans cette section , nous allons dmontrer le thorme [ theo2 ] .
nous allons dans un premier temps noncer plus prcisment lassertion en question , puis nous en crirons la preuve .
nous verrons quelle consiste essentiellement en une analyse raffine du lemme prcdent [ lem : cle ] , en effet nous ferons ce quon pourrait appeler un `` contrle de la monte de lordre '' par rapport la forme @xmath350 .
+ rappelons que dans ce paragraphe , @xmath240 gale @xmath215 .
* soient @xmath351 deux formes non nulles de @xmath48 .
ecrivons @xmath352 avec @xmath353 .
on dit que @xmath71 est infrieure ( resp .
strictement infrieure ) @xmath72 si @xmath354 ( resp .
@xmath355 ) . on abrgera
cette notion en notant @xmath356 ( resp .
@xmath357 ) .
par convention @xmath358 , toute forme @xmath50 est infrieure @xmath359 . *
soient @xmath360 dans @xmath48 et @xmath361 .
on dit que @xmath328 est @xmath71-homogne si @xmath362 .
on dit que @xmath328 est @xmath363-homogne si @xmath364 .
* soit @xmath50 une forme dans @xmath48 .
nous noterons @xmath365 lordre sur @xmath366 donn par : @xmath367 @xmath368 nous remarquons quen adoptant les notations du paragraphe prcdent et en posant @xmath369 ( avec @xmath370 ) alors on a : @xmath371 .
si par contre @xmath372 alors @xmath373 .
soit @xmath63 un cne de @xmath103 de dimension @xmath215 ( maximale ) et @xmath374 avec @xmath375 .
notons @xmath171 la base standard de @xmath99 associe @xmath63 .
on dfinit @xmath376 par : @xmath377 avec les notations prcdentes , nous avons ( voir la figure [ fig : kappa1 ] ) : @xmath378 soit @xmath257 un
cne de dimension maximale et soient @xmath383 ses gnrateurs primitifs .
soit @xmath384 avec @xmath385
dans @xmath386 , en particulier @xmath387 .
supposons donn @xmath388 tel que @xmath389 et @xmath390 et tel que @xmath391 .
alors nous avons montr dans le lemme [ lem : cle ] comment construire , en un nombre fini dtapes , un lment @xmath392 tel que @xmath393 ( i.e. lordre par rapport @xmath71 na pas augment ) et @xmath394 ( i.e. lordre par rapport
@xmath72 a baiss le plus possible ) .
nous pouvons nous demander ce qui se passe pour lordre @xmath350 de @xmath392 par rapport celui de @xmath113 .
nous allons montrer que cet ordre peut augmenter mais de manire contrle .
cest lobjet du lemme suivant : [ lem : controlev1 ] soit @xmath63 un cne de dimension maximale de @xmath214 et @xmath395 ses gnrateurs primitifs ( il est possible que @xmath396 ) .
supposons @xmath397 .
+ soient @xmath398 et @xmath399 .
soit @xmath400 tel que @xmath389 et @xmath390 alors on peut construire @xmath401 partir de @xmath113 tel que : par hypothse il existe @xmath411 vrifiant @xmath412 et @xmath413 . on dfinit @xmath414 ( il existe des entiers @xmath415 , @xmath416 , et @xmath417 satisfaisant une telle galit ) , @xmath303 et @xmath418 .
on reprend le dbut de la preuve du lemme [ lem : cle ] la diffrence quon travaille avec la forme @xmath72 au lieu de @xmath71 .
on considre donc la division de @xmath308 par la base standard @xmath171 relativement lordre
@xmath419 ce qui donne : + @xmath420 avec @xmath421 . on note
@xmath153 lensemble des @xmath203 dans @xmath422 pour lesquels lingalit prcdente est une galit . on pose alors : dmontrons ces affirmations : + * ( a ) : * nous avons
@xmath428 et ce dernier est major par @xmath429 donc si * ( b ) * est vrai il en est de mme pour * ( a)*. + * ( b ) : * comme au paragraphe prcdent , on peut montrer que la division de @xmath430 par @xmath431 relativement @xmath419 donne : @xmath423 .
par consquent , il existe @xmath432 tel que @xmath433 o @xmath434 qui est un monne de @xmath435 .
en particulier , ceci implique : @xmath436 dautre part , @xmath437 soit alors @xmath438 tel que @xmath439 . en prenant @xmath440 , on obtient @xmath441 remarquons quil
est possible quon ait @xmath442 .
cependant , on a toujours : ( voir la figure [ fig : affirmations ] ) en utilisant cette affirmation et les identits ( [ ident5 ] ) et ( [ ident6 ] ) , nous obtenons : @xmath444 ceci dmontre le point * ( b)*. il ne reste plus qu dmontrer le point * ( c)*. + la division de @xmath445 par @xmath446 relativement @xmath419 donne le rsultat suivant : @xmath447 .
par consquent , @xmath448 or @xmath449 et @xmath450 donc @xmath451 or @xmath452 donc @xmath453 le point * ( c ) * est dmontr .
+ voyons maintenant comment laffirmation * ( a ) * permet de montrer le troisime point du lemme .
nous sommes partis de @xmath328 et nous avons construit @xmath322 .
par * ( a ) * , nous avons @xmath454 .
la suite consiste faire les mmes oprations avec @xmath455 la place de @xmath328 .
le dernier lment @xmath456 ainsi construit vrifie : @xmath457 et @xmath458 .
+ on pose alors @xmath459 , on a bien @xmath460 et @xmath461 .
le lemme est dmontr .
notons @xmath462 les lments primitifs du @xmath66-squelette de @xmath103 .
pour chaque @xmath276 , notons @xmath463 le cne contenant le cne ouvert engendr par @xmath464 et @xmath246 .
+ soit @xmath465 .
+ montrons par rcurrence sur @xmath329 que pour tout @xmath466 , il existe @xmath467 vrifiant : pour @xmath471 : @xmath472 donc en particulier @xmath473 ( notons
que @xmath474 ) donc il existe @xmath475 tel que @xmath476 et @xmath477 .
+ supposons lassertion vraie au rang @xmath478 .
+ on applique le lemme [ lem : controlev1 ] avec @xmath479 et @xmath480 . on pose alors
@xmath481 ( notations du lemme ) .
daprs ce mme lemme , @xmath482 vrifie : le procd de construction de @xmath489 montre quun tel @xmath490 nest priori pas unique . en effet ,
en inversant les rles de @xmath350 et @xmath359 , on aurait pu construire un @xmath490 de la forme @xmath491 .
sab87b a. assi , f.j .
castro jimnez , m. granger , _ the analytic standard fan of a @xmath492-module _ , journal of pure and applied algebra * 164 * ( 2001 ) , 3 - 21 .
r. bahloul , _ contributions ltude des idaux de bernstein - sato dun point de vue constructif _ , thse de doctorat ( 2003 ) , universit dangers . i. n. bernstein , _ the analytic continuation of generalised functions with respect to a parameter _ , funct .
* 6 * ( 1972 ) , 273 - 285 .
bjrk , _ dimensions over algebras of differential operators _ ( 1973 ) , preprint .
castro jimnez , l. nrvaez - macarro , _ homogenising differential operators _ , preprint n@xmath493 * 36 * ( 1997 ) , universidad de sevilla .
a. gyoja , _ bernstein - sato s polynomial for several analytic functions _ , j. math .
kyoto univ .
* 33*(2 ) ( 1993 ) , 399 - 411 .
m. kashiwara , _ b - functions and holonomic systems .
rationality of roots of b - functions _ , invent . math . * 38*(1 ) ( 1976 ) 33 - 53 .
b. lichtin , _ generalized dirichlet series and @xmath494-functions _ , compositio math * 65 * ( 1988 ) , 81 - 120
. b. malgrange , _ le polynme de bernstein dune singularit isole _ , lecture notes in math . * 459 * ( 1975 ) , springer verlag , 98 - 119 . c. sabbah , _ proximit vanescente i. la structure polaire dun @xmath492-module _ , appendice en collaboration avec f.j .
castro jimnez , compositio math .
* 62 * ( 1987 ) , 283 - 328 . c. sabbah , _ proximit vanescente ii .
equations fonctionnelles pour plusieurs fonctions _ , compositio math . * 64 * ( 1987 ) , 213 - 241 . | in 1987 , c. sabbah proved the existence of bernstein - sato polynomials associated with several analytic functions . the purpose of this article is to give a more elementary and constructive proof of the result of c. sabbah based on the notion of the analytic grbner fan of a @xmath0-module . |
the synthetic field method ( sfm ) was developed by gonzlez et al .
( 1998 ) as a technique for determining the average opacity through the disk of a nearby spiral by counting more distant background galaxies .
this number is compared with those of `` synthetic fields '' , a known wfpc2 deep field , dimmed by a certain opacity , added to the foreground galaxy image .
gonzlez et al . ( 2003 ) examined the application of the method and concluded that the optimum results with hst / wfpc2 imaging would be obtained on fornax and virgo cluster galaxies , with the accuracy degrading for galaxies much closer or more distant .
our sample was drawn from the hst archives , in part from the cepheid distance scale hst key project sample .
we present preliminary results on ngc4535 , ngc4725 and ugc2302 . the numbers of real and simulated field galaxies were compared in annuli of deprojected radius .
ngc4535 is a face - on sabc with a half - light radius of 5.1 kpc at a distance of 16 mpc .
several spiral arms with bright hii regions are visible in the hst image .
ngc4725 is a sabab ringed galaxy at 12 mpc its half - light radius is 5.4 kpc .
the hst image shows much less structure .
the radial opacity profiles of both however are similar ( see figure 1 ) , showing 2.5 magnitudes of extinction in i at the half - light radius , dropping off to 0.5 magnitude .
ugc2302 is a lsb at 15 mpc is a much more compact object . at comparable radii
, it does not show as much absorption ; less then a magnitude at 5 kpc from the center .
the uncertainties in these individual profiles are large due to low numbers of field galaxies and an added uncertainty due to their clustering . by averaging a number of fields we will improve on these uncertainties for an average radial opacity model of spiral galaxies . for different hubble types , we intend to compare the radial dependence of opacity with the hi density profile to obtain cold dust - to - gas ratios .
gonz ' alez , r. a. , allen , r. j. , dirsch , b. , ferguson , h. c. , calzetti , d. , and panagia , n. 1998 , * 506 * , 152 gonz ' alez , r. a. , loinard , l. , allen , r. j. , and muller , s. 2003 , * 125 * , 1182 holwerda , b. w. , allen , r. j. , and van der kruit , p. c. : 2002 , asp conf .
ser . 273 : 337 | we have applied the `` synthetic field method '' on a sample of 20 nearby galaxies in order to determine the opacity of their disks .
we present preliminary results on the radial dependence of cold dust absorption for 3 examples .
the spirals ngc4535 and ngc4725 show significant absorption at a half - light radius . ugc2302 , a lsb galaxy , shows much less opacity . # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
the primary mission of the batse instrument was to detect and locate @xmath0-ray bursts .
while the 8 detectors had no intrinsic positional sensitivity , burst positions were obtained by triangulation between detectors .
it was realised , prior to launch , that persistent sources of @xmath0-rays could also be observed by using the earth occultation technique @xcite .
this method calculates the source flux by measuring the step in the count rate of each batse detector as a source rises or sets below the earth s limb .
mass modelling is a technique to simulate the background radiation experienced by a space craft @xcite .
the batse mass model ( bamm ) is a geant3 monte - carlo simulation code developed at southampton @xcite .
bamm simulates the expected count rate from cosmic diffuse @xmath0-rays , atmospheric albedo @xmath0-rays and cosmic rays thoughout the orbit .
the position and orientation of the spacecraft relative to the earth are used to filter these results and generate the expected background .
bamm has been used to flat - field the entire 9 year batse data set .
it is thought that the origin of the exceptionally high luminosity of active galaxies is the accretion of matter onto a supermassive black hole located at the galaxy centre @xcite .
agn typically have highly variable x - ray and @xmath0-ray fluxes .
the very short timescale variability indicates that this emission originates from a small region very close to the central engine .
hence the variability of agn yields direct information about the central supermassive black hole .
the crab is one of the brightest @xmath0-ray sources in the sky and is well known for its lack of variability . as
such it is the ideal ` standard candle ' with which to test the flat - fielded data set and explore the capabilities of our methods .
[ fig : crab ] shows the 9 year light curve of the crab in the 30 - 230 kev energy band as measured on a daily basis and with a 53 day moving average fit ; 53 days is the precession period of cgro .
a histogram of this light curve is seen in fig .
[ fig : crabhist ] .
the crab appears relatively constant in the light curve , hence the histogram should have a gaussian profile . fitting a gaussian distribution to the histogram
yields a centroid of 1.005 @xmath1 0.002 crabs with a reduced @xmath2@xmath3 of @xmath43 .
examining the residuals of this fit indicates a number of spurious points in the 0.6 - 0.9 crab region which may be the result of an unknown low level systematic , removing these points renders a fit with a @xmath2@xmath3 of 1.3 .
rebinning the data into longer timescale averages , before generating the histogram , allows the estimation of the systematic component of any errors . as the time bins increase in size the standard deviation of the data set would ideally asymptotically approach 0 .
the inset graph of fig .
[ fig : crabhist ] shows that the curve approaches @xmath42.5% , indicating this level of systematic error .
this level is assumed in our studies of agn light curves .
agn are some of the most powerful persistent sources of @xmath0-rays in the observable universe .
however the low fluxes which arrive at earth make them a very challenging subject to observe in this waveband .
the three brightest agn seen in batse are : centaurus a , a radio galaxy ; ngc4151 , a seyfert galaxy ; 3c273 , a blazar . fig .
[ fig : lightcurves ] shows the light curves in the batse 20 - 50 kev and 50 - 160 kev band of these agn .
the data have been binned up to 30 day data points in order to minimise any errors .
additionally the rxte - asm 2 - 10 kev light curves are presented , scaled and offset from the batse curves to aid legibility .
[ fig : hard - soft ] is a hard flux - soft flux plot of the three agn showing the flux in the 30 - 50 kev band against that of the 50 - 160 kev band .
the hard and soft batse light curves seen in the upper panel of fig .
[ fig : lightcurves ] appear to be well correlated apart from a short , @xmath430 day outburst in the 20 - 50 kev band at tjd @xmath410800 which appears to coincide with a drop in the 50 - 160 kev flux .
however , the 20 - 50 kev flux continues to be correlated with the asm light curve during this period .
this appears to be corroborated by the hard - soft plot where the hard and soft fluxes indicate a positive correlation .
a spearman rank correlation coefficient of 0.56 with a probability of chance occurance of [email protected] confirms this positive correlation .
the asm light curve follows the general trends exhibited by both of the batse light curves neglecting the outburst at tjd @xmath410800 .
the hard and soft light curves shown in the middle panel of fig . [ fig : lightcurves ] appear to be well correlated although there appears to be a potential time lag between them of @xmath430 days .
emission in the 50 - 160 kev band initially appears to precede that of the 20 - 50 kev band , however this is not obvious after tjd @xmath48800 . as this lag is on the same time scale as the data binning it is tenuous without a detailed statistical analysis .
the hard - soft plot looks flat as the hard flux appears independent of the soft flux , however , a positive correlation is indicated by a spearman rank correlation coefficient of 0.31 with a probability of chance occurance of [email protected] . both light curves seen in the lower panel of fig .
[ fig : lightcurves ] are synchronized and well correlated .
they both show the same general trends .
the hard - soft plot indicates a positive correlation with a spearman rank correlation coefficient of 0.45 with a probability of chance occurance of [email protected] . ]
the 9 year batse data set has been flat - fielded using bamm to remove temporal variations in the instrument background .
we have then applied the standard earth occultation technique developed at marshall space flight centre @xcite on the flat - fielded data set and made measurements to the crab nebula and 3 agn : cen a ; ngc4151 ; 3c273 .
the crab light curve exhibits a constant flux of 1.005 @xmath1 0.002 crabs confirming that the flat - fielding had no adverse effect on the data .
additionally , the systematic error of the methods and analysis performed in the generation of the light curve is estimated to be limited to @xmath42.5% .
the agn light curves show that batse is sensitive to the variations in their @xmath0-ray fluxes .
work is in progress to re - optimise the standard earth occultation technique for use with the flat - fielded data with the intention of improving sensitivity and precision of measurements .
we are simultaneously using the flat - fielded data with new methods to generate all - sky images for the whole 9 years @xcite . , s. e. , bird , a. j. , dean , a. j. , et al .
, 2001 , in exploring the gamma - ray universe : proceedings of the 4th integral workshop , alicante , 2000 , ed
. a. gimenez , v. reglero , & c. winkler , vol .
esa sp-459 , 521524 | the burst and transient source experiment ( batse ) aboard cgro monitored the whole sky through 8 nai(tl ) crystal scintillators sensitive in the 20 kev - 2 mev energy band continuously from april 1991 until june 2000 .
results are presented on the long term variabilty observed in the brightest active galactic nuclei ( agn ) present in the batse data archive .
this was achieved through the application of the earth occultation technique to data flat - fielded using the batse mass model .
removal of the temporal background variations from the data should allow a more sensitive extraction of source parameters .
analysis of the general trends of the 9-year light curves are also presented .
[ 2004/02/13 1.1 ( pwd ) ] |
in the conventional quark model , a baryon is a bound state of three more - or - less equivalent valence quarks .
however , there are variants of the quark model in which interactions give rise to a quark - diquark structure for the baryon , where the binding of two of the quarks in the baryon is different from the bound state properties of the third quark . for a review of diquark phenomenology ,
see @xcite .
one would like to understand the properties of diquark correlations ( if any ) directly from qcd . since this is a problem of the long distance behavior of the strong interactions , perturbation theory is not applicable .
lattice qcd is a potential source of information about diquark structures , and the subject has been visited many times in lattice simulations @xcite .
diquarks are not color singlets .
many techniques have been used in lattice simulations to deal with this fact . in refs .
@xcite , gauge dependent diquark correlators were calculated in a fixed gauge to study the masses of diquarks . in refs .
@xcite gauge invariance was preserved by investigating the properties of states containing a static quark coupled gauge - invariantly to two light quarks .
their quantum numbers combine to give the quantum numbers of the diquark . whatever a diquark is , the environment it feels in a baryon with a heavy spectator quark is presumably different from that in a baryon with a third light quark . in this work ,
we want to study diquark correlations in light baryons .
the interaction between a diquark and a light quark probably depends on the spin of the diquark and this interaction can affect the formation of the diquark in the light baryon .
here we combine a light quark with a diquark with a specific quantum number in the color anti - triplet representation to get a color singlet .
this work addresses two questions : first , do baryon correlators involving light quarks show qualitative features associated with diquarks ?
we attack this question using both quenched simulations and data from simulations with two flavors of light dynamical fermions . as the reader will see , we had to deal with a second , different question along the way : are the properties of baryon correlators affected by the presence of dynamical fermions ?
the answer to that question is yes : with the simulation parameters we used , the quenched approximation is afflicted by artifacts which make it unsuitable to look for diquark effects .
we look for diquark effects through the point to point correlator for baryonic currents , using the methodology first described by shuryak @xcite : we compare the correlator to its free - field value .
if the ratio of the full correlator to its free field value exceeds unity , we say that the channel is attractive ; if it falls below unity we say that the channel is repulsive .
an unphysical situation which is encountered in quenched simulations is that the ratio becomes negative .
this indicates a violation of unitarity .
we are interested in diquarks in the color anti - triplet representation since diquarks in the color sextet representation have much larger color electrostatic energy and thus are not favored by phenomenology @xcite .
the scalar diquark is often called a good " diquark since the two quarks are thought to be attractive to each other . the @xmath2 diquark is called a bad " diquark because all models suggest that it is heavier than the scalar diquark .
besides these two diquarks , we also study the pseudoscalar diquark . in a quenched simulation , the point to point scalar meson correlator was found to be negative @xcite .
this behavior , and similar behavior for the scalar correlator in other situations , is known to be a quenching artifact . since quark - quark interactions resemble quark - antiquark interactions , we are curious to see if we will see similar quenching artifacts in the pseudoscalar diquark channel .
excess zero modes of dirac operators at small quark masses are lattice artifacts in quenched lattice simulations . in a preliminary version of this work , reported in @xcite
, we found that behavior of baryon correlators was strongly affected by zero modes . in the sector of our data which had no zero modes ,
the difference among different baryon correlators was much smaller . in that study , we also had data from simulations with dynamical fermions , but they were performed in sectors of fixed topology and could not be applied directly to phenomenology
. here we will give results from dynamical simulations in which topological charge is allowed to fluctuate .
the paper is organized as follows : in section [ meth ] , we present our set up and simulation details . numerical results and discussions
are given in section [ numres ] .
then we conclude in section [ summ ] .
our simulations are performed using overlap fermions .
we have two data sets : one is quenched and the other is from a simulation with two degenerate flavors of light fermions .
the simulations have many features in common , allowing for a reasonably straightforward comparison of results . in both simulations
the `` kernel '' ( nonchiral dirac operator @xmath3 inserted into the overlap operator @xmath4 ) contains nearest and next - nearest terms and a clover term set to its tree level value .
the actual couplings are slightly different ; the action used in the quenched simulations is described in ref .
@xcite while the dynamical action is given in ref .
@xcite .
both simulations use gauge connections which were blocked over a hypercube .
the quenched simulations employ the implementation of hasenfratz and knechtli @xcite while the dynamical simulations used the differentiable hypercubic link of hasenfratz , hoffmann , and schaefer @xcite . the choice of blocking parameters , @xmath5 , @xmath6 , @xmath7 , in the conventions of ref .
@xcite , are identical .
these two blockings produce identical results in perturbation theory .
our quenched data set has 40 configurations with lattice size @xmath8 and were generated using the wilson gauge action at a gauge coupling @xmath9 .
the bare quark masses @xmath10 are 0.015 , 0.025 and 0.05 .
the lattice spacing is @xmath11fm determined from the sommer parameter @xmath12 , assuming a nominal physical value of 0.5 fm .
thus the lattices have a physical size of about 1.3 fm , and the quark masses are about 37 , 60 , and 120 mev . the quenched simulations exhibit a typical quenched topological susceptibility .
from the zero - mode spectrum , they have @xmath13 and @xmath14 .
this corresponds to a scaled topological susceptibility of @xmath15 or @xmath16 mev)@xmath17 , not too different from recent high - statistics results @xcite . in the two flavor dynamical simulation
the lattice size is @xmath18 .
we used a tadpole - improved lscher - weisz action with a coupling @xmath19 and performed simulations at three quark masses : @xmath20 , 0.05 and 0.10 .
the sommer parameters are found to be @xmath21 at quark mass 0.03 , 3.49(4 ) at 0.05 and 3.39(3 ) at 0.10 .
thus the lattice spacing is about 0.14 fm , the physical size of these lattices is @xmath22 fm , and the quark masses are 43 , 69 , and 133 mev .
the pion masses in lattice units are 0.32 , 0.43 and 0.59 respectively .
details of the simulations are found in ref .
our measurements of the correlators are done on 60 to 80 configurations from these data sets .
the topological susceptibility shows a dramatic variation with the quark mass , roughly consistent with the expected linear dependence @xmath23 . a strength of our simulation is the use of a fermion with continuum - like chiral symmetry at finite lattice spacing .
it will turn out that the correlation functions we measure have a fair amount of sensitivity to the underlying topology of the gauge configurations . simply counting the number of chiral zero modes
gives us a robust measure of the topology .
another strength is the use of full qcd with identical valence and sea quarks , rather than using simulations with different formulations of valence and sea quarks , or partially quenched simulations ( i.e. , with different valence and sea quark masses ) .
this way we avoid a host of artifacts associated with the peculiar chiral properties of such formulations
. the particular disease we want to avoid is associated with scalar correlators .
this is best known in the situation of scalar mesons , where there is a long history of observations going back to the quenched studies of ref .
basically , the scalar meson correlator includes a pi - eta - prime intermediate state . in
quenched or partially quenched simulations , the flavor singlet pseudoscalar channel contains an unphysical double pole at finite lattice spacing .
this intermediate state can dominate the correlation function ; removing it requires all the machinery of partially quenched chiral perturbation theory ( see ref .
@xcite , plus the recent summaries of refs .
@xcite ) . in addition
, zero modes of the dirac operator ( in sectors of nonzero topology ) make a negative contribution to the scalar correlator @xcite .
this has been seen in point - to - point meson correlators in small volume @xcite .
the current algebra description for scalar mesons is of course lacking for pseudoscalar combinations of two quarks , but the disease is still present .
we will see negative correlation functions in our quenched data .
they are expected to be absent in our full qcd simulations , and they will be .
an obvious weakness of our study is that the simulation volume is small .
we do not trust our measurements of baryon masses ; while we can use a combination of quark propagators which are periodic and antiperiodic in time to go to long temporal separations in the correlation function , the three - dimensional volumes probably squeeze the baryon wave functions , affecting their masses . in addition , as the quark mass falls to zero at fixed volume @xmath24 , at some point @xmath25 becomes order unity . at this point
, the finite volume induces an effective restoration of chiral symmetry ; the condensate is pushed to zero .
this so - called epsilon - regime is interesting in its own right , but can not be used to make direct connections to the properties of hadrons in infinite volume .
clearly , any future simulation should be done with larger volume .
the baryon currents and correlation functions we considered are collected in table [ currenttab ] . here
@xmath26 is the charge - conjugation operator .
.currents and correlation functions [ cols="^,^,<,<",options="header " , ] [ currenttab ] both the current @xmath27 and @xmath28 contain a scalar ( good " ) diquark . if the scalar diquark correlation is important in the correlators for both currents , we expect to see similar behavior in both correlators .
the currents @xmath29 and @xmath30 contain a pseudoscalar diquark and an axial vector ( bad " ) diquark respectively .
under exchanges of all indices for space , dirac , flavor and color , diquark structures should be antisymmetric .
since we are considering two quarks in the color anti - triplet representation , they are antisymmetric with respect to the color index . as for the space index ,
they are symmetric .
thus the symmetric property of @xmath31 ( @xmath32 , @xmath1 , @xmath33 and @xmath34 ) determines the symmetric properties of the diquarks under the exchange of the flavor index . for @xmath32 , @xmath1 and @xmath33 ,
the diquark states are antisymmetric under the exchange of dirac indices .
therefore they have to be flavor isoscalars . for @xmath35 ,
the diquark state is a flavor isovector .
the two point correlator for a current @xmath36 is defined as @xmath37 , where @xmath38 is the vacuum and @xmath39 is the time order operator .
for free massless quarks , the euclidean coordinate space quark propagator takes the form @xmath40 the index @xmath41 is summed over . for the current @xmath27 in table [ currenttab ]
, we have @xmath42 similarly , we can get the free correlators for the other currents in table [ currenttab ] .
they are the same as the result in eq .
( [ j5free ] ) except for a sign flip for @xmath28 and @xmath43 ( for @xmath44 , @xmath45 is needed to get the same result ) . as was done in refs .
@xcite , to reduce the sensitivity on quark masses of the free correlators , which we will use to normalize the interacting correlators , it is convenient to multiply the correlators with @xmath46 and take the trace in the dirac indices for all the currents .
for example , from eq.([j5free ] ) we find @xmath47=-15/4\pi^6x^8 $ ] .
@xmath48 is used to normalize the interacting correlator @xmath49 , i.e. we will examine the ratio @xmath50 for each current . in our lattice simulations , we use the free lattice correlators @xmath48 , rather than an analytic formula , to do the normalization to reduce lattice artifacts . because of the periodic boundary conditions in spatial directions , the correlators receive contributions from image points of the source , in addition to their direct " contribution .
when computing the free correlator , we should only keep contributions , where the source points for the three quarks coincide , and so do the three sink points .
they yield the infinite volume correlator as we want .
the contributions from the cross terms are finite volume artifacts . in these cross terms , the three quark sources sit on different image points and the three sink points coincide .
we expect that these cross terms affect the free and interacting correlators differently . for the quenched system , the cross terms do not contribute since they correspond to polyakov loops encircling the entire lattice and confinement forces their contributions to be zero .
for the system with dynamical fermions , propagation of different quarks to different image points involves intermediate states in which the quarks separate a distance @xmath51 apart . presumably , these contributions are @xmath52 effects .
however for the free theory , the cross terms give a nonzero contribution to the correlators , which is power law in @xmath51 , since there is no confinement .
we can not eliminate the image contributions from the free propagator .
however , we can suppress them .
we do this via the method of ref @xcite : the free overlap fermion propagators are calculated on a larger lattice ( @xmath53 ) than the lattices for the interacting simulations . then the correlators in the small volume are approximated as a sum of a term from the source and a term from the nearest image point . on the lattice , correlation functions in the coordinate space for mesons , nucleon and @xmath54 were calculated in refs .
@xcite . in @xcite ,
the correlators were fit to functions from the corresponding parametrization of the spectral density functions to an isolated resonance contribution and a continuum contribution .
here we focus on the effects of diquark correlations on baryon correlators .
we do not try to do any fit of the correlators because our lattice size is small .
data at different values of @xmath55 are , of course , highly correlated , because they come from the same set of underlying configurations .
the statistical errors which we show in our plots are from a jackknife analysis .
fig . [ free_r0x ] shows the continuum and lattice results for the free correlator @xmath48 of the current @xmath27 .
we show one graph for each of the quenched and dynamical simulations because the parameters for the two kernel " dirac operators of our overlap dirac operator are slightly different . for @xmath27 from a @xmath53 lattice .
the solid line is the continuum result @xmath56 .
the left graph is for the lattice action used in the quenched simulation .
the right for the lattice action used in the dynamical simulation .
, title="fig:",width=188,height=188 ] for @xmath27 from a @xmath53 lattice .
the solid line is the continuum result @xmath56 .
the left graph is for the lattice action used in the quenched simulation .
the right for the lattice action used in the dynamical simulation .
, title="fig:",width=188,height=188 ] the lattice results agree quite well with the continuum result except at very small distance , where lattice discretization artifacts are apparent . this behavior is seen for all the free correlators we calculated . in the following
, we will restrict our discussion of the results in the region @xmath57 ( quenched ) or @xmath58 ( dynamical ) according to the lattice size .
we expect that the anisotropy introduced by the lattice will cancel out to some extent in the ratio @xmath50 .
[ quen_all ] shows the normalized correlation functions @xmath50 against the physical distance @xmath55 for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 for three quark masses .
the first graph shows that the correlator for @xmath27 is attractive ( @xmath60 ) and the attraction increases as the quark mass decreases .
for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from the quenched simulations for three quark masses.,title="fig:",width=188,height=188 ] for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from the quenched simulations for three quark masses.,title="fig:",width=188,height=188 ] + for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from the quenched simulations for three quark masses.,title="fig:",width=188,height=188 ] for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from the quenched simulations for three quark masses.,title="fig:",width=188,height=188 ] for @xmath28 , at the largest quark mass the correlation function is repulsive ( @xmath61 ) , which is very different from the correlator for @xmath62 although both of them contain a scalar diquark . as the quark mass decreases , the correlator for @xmath28 becomes flat and then seems to curve up , showing some attraction .
for the current @xmath43 , which contains a pseudoscalar diquark , we find that at the smallest quark mass the normalized correlation function tends to go negative at large distance .
this behavior is very similar to what was observed for the scalar meson in ref .
it was argued in ref .
@xcite that zero modes are the sources of the negativity . in the following , we will see zero modes do make a big difference here .
the last graph in fig .
[ quen_all ] is for the current @xmath59 , which contains an axial vector ( bad " ) diquark .
some attraction shows up above @xmath63 fm and seems to increase as the quark mass decreases .
comparing with @xmath27 , the attraction in this channel is much weaker .
for the four currents for the lowest quark mass , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] for the four currents for the lowest quark mass , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] + for the four currents for the lowest quark mass , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] for the four currents for the lowest quark mass , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] to see zero mode effects , we separate the 40 configurations into two groups : 35 of them with topological charge @xmath64 and the other 5 with @xmath65 .
i.e. in the first group the quark propagator contains zero modes while in the second one it does not . from each group
we compute the correlation functions and compare .
the results are shown in fig .
[ quen_q0.015 ] for our smallest quark mass and fig .
[ quen_q0.050 ] for the largest quark mass . for the current @xmath27 , at the lowest quark mass the attraction at large @xmath55 ( @xmath66 fm ) is larger in the @xmath67 sector . for @xmath28 ,
the zero mode effects are in the opposite direction of those for @xmath62 .
the correlator from the @xmath65 group ( no zero mode ) is greater than the correlator from the @xmath64 group .
i.e. , zero modes make the @xmath28 correlator less attractive . for @xmath29 ,
the zero modes contribute strongly to the repulsion seen at the lowest quark mass 0.015 so that @xmath50 becomes negative at large @xmath55 . for @xmath30 , when we compare the correlators from the @xmath64 group and from the @xmath65 group , we find no difference within error bars .
zero mode effects are not important in this channel . in fig .
[ quen_q0.050 ] , we see similar , but smaller , zero mode effects in each channel for @xmath68 .
this is natural given how zero modes scale with quark mass .
for the four currents for @xmath69 , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] for the four currents for @xmath69 , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] + for the four currents for @xmath69 , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] for the four currents for @xmath69 , from quenched data .
we show the comparison of the correlators from configurations with @xmath64 and from configurations with @xmath65.,title="fig:",width=188,height=188 ] if we only use the @xmath65 configurations from our quenched simulation , we obtain the correlation functions shown in fig .
[ quen_all_q0 ] . for the four currents computed from the configurations with @xmath65 for the three quark masses.,title="fig:",width=188,height=188 ] for the four currents computed from the configurations with @xmath65 for the three quark masses.,title="fig:",width=188,height=188 ] + for the four currents computed from the configurations with @xmath65 for the three quark masses.,title="fig:",width=188,height=188 ] for the four currents computed from the configurations with @xmath65 for the three quark masses.,title="fig:",width=188,height=188 ] comparing with fig .
[ quen_all ] , we see that in the @xmath27 channel the attraction is weaker and its dependence on the quark mass is also weaker .
the @xmath28 channel shows a little bit more attraction than in fig .
[ quen_all ] at the smallest quark mass @xmath70 .
the biggest change is in the @xmath43 channel .
the @xmath43 correlators for all three quark masses are attractive in fig .
[ quen_all_q0 ] .
in contrast , in fig .
[ quen_all ] , they are repulsive and even negative . for @xmath59 , there is no difference within error bars between fig .
[ quen_all ] and fig .
[ quen_all_q0 ] since zero mode effects are small in this channel . from the quenched simulation for the lowest quark mass .
the left graph is from all configurations .
the right from the configurations with @xmath65.,title="fig:",width=188,height=188 ] from the quenched simulation for the lowest quark mass .
the left graph is from all configurations .
the right from the configurations with @xmath65.,title="fig:",width=188,height=188 ] to summarize the quenched results , we find that the attraction in the @xmath27 channel , which contains a scalar diquark structure , is the strongest .
the @xmath43 channel is repulsive and even becomes negative at large distance as the quark mass decreases .
the @xmath28 channel is repulsive at the large quark mass , but turns into attractive at small quark masses .
its behavior is quite different from the correlator for @xmath27 even though @xmath28 also contains a scalar diquark .
the axial vector channel is weakly attractive and the attraction seems to increase when we go to light quark masses .
the comparison of the four correlators obtained from all 40 configurations at the lowest quark mass @xmath70 is shown in the left graph in fig .
[ quen0.015 ] .
the @xmath27 channel is very different from the other channels . by comparing the correlators from different topological groups
, we find that at small quark masses both the attraction in the scalar diquark channel ( @xmath27 ) and the repulsion in the pseudoscalar diquark channel ( @xmath43 ) have big contributions from configurations with zero modes . in the @xmath28 channel ,
the zero modes make the correlator a little less attractive . in the right graph of fig .
[ quen0.015 ] , we show the comparison of the four correlators obtained from configurations without zero modes at the lowest quark mass 0.015 . the difference between channel @xmath27 and @xmath43 shrinks considerably .
the attractions in these two channels are similar and stronger than the attractions in @xmath28 and @xmath59 . in full qcd simulations , configurations with high @xmath71 values
are suppressed by the fermion determinant at small quark masses .
but certainly not all configurations are with @xmath65 .
thus we expect to see a picture somewhere between the two graphs in fig .
[ quen0.015 ] at small quark masses in our dynamical simulation . in fig .
[ dyn ] , we show the normalized correlation functions for all four currents from our simulations with two flavors of dynamical fermions . for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from two flavor dynamical simulations for three quark masses.,title="fig:",width=188,height=188 ] for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from two flavor dynamical simulations for three quark masses.,title="fig:",width=188,height=188 ] + for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from two flavor dynamical simulations for three quark masses.,title="fig:",width=188,height=188 ] for the four currents @xmath27 , @xmath28 , @xmath43 and @xmath59 from two flavor dynamical simulations for three quark masses.,title="fig:",width=188,height=188 ] in the correlator for @xmath27 , we see an attraction which increases as the quark mass decreases .
this attraction is weaker and increases more slowly than in the quenched data set , in fig .
[ quen_all ] .
the @xmath28 correlator is repulsive at @xmath72 .
but as the quark mass decreases , it becomes less repulsive and at @xmath20 , it is unity in the range of @xmath55 that we consider .
this mass dependence is consistent with the quenched simulation result ( see the second graph in fig .
[ quen_all ] ) .
the dynamical result confirms that the @xmath28 correlator behaves quite differently from the @xmath62 correlator although both currents contain a scalar diquark .
the @xmath43 correlator does not become negative at smaller quark masses ; instead , it becomes more attractive .
we saw similar behavior in the @xmath65 sector of the quenched data set . here
full qcd is unitary and the correlator remains positive .
the current @xmath43 vanishes in the nonrelativistic limit .
thus it is not surprising that the @xmath43 correlator is repulsive at big quark masses .
the attraction at small quark masses is a relativistic effect and we expect it to become stronger as the quark mass becomes smaller . the @xmath59 correlator , which involves the bad " diquark , is repulsive at the largest quark mass , @xmath72 .
then as the quark mass decreases , it becomes slightly attractive .
as in the quenched simulation , the attraction in this correlator is much weaker than in the @xmath27 correlator . in fig .
[ dyn_4j ] , we put all four correlators together for comparison .
the left graph is for the largest quark mass @xmath72 .
the right graph is for the lowest quark mass @xmath20 . from the dynamical simulation .
the left graph is for the largest quark mass .
the right for the lowest quark mass.,title="fig:",width=188,height=188 ] from the dynamical simulation .
the left graph is for the largest quark mass .
the right for the lowest quark mass.,title="fig:",width=188,height=188 ] when the quark mass is big , only the @xmath27 correlator is attractive .
the other three are repulsive as we can see in the left graph in fig .
[ dyn_4j ] .
as the quark mass becomes small in the dynamical simulation , the other three correlators turn into attractive as is shown in the right graph in fig .
[ dyn_4j ] .
however the attraction in the @xmath27 correlator is still much stronger than in the other three .
we also observe a slightly stronger attraction in the @xmath43 correlator than in the @xmath28 and @xmath59 correlators . , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath20 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath20 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] + , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath20 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath20 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath74 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath74 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] + , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath74 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath74 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath72 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath72 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] + , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath72 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] , @xmath28 , @xmath43 and @xmath59 from @xmath73 simulations at quark mass @xmath72 .
the correlators from configurations with @xmath64 are compared with those from configurations with @xmath65.,title="fig:",width=188,height=188 ] as we did in the analysis of the quenched data , we can separate our dynamical configurations according to their topological charges .
we have 60 configurations for @xmath72 and 80 configurations for both @xmath74 and 0.03 .
the numbers of high @xmath71 ( @xmath75 ) configurations for the three quark masses are 37 , 33 and 16 respectively .
clearly , high @xmath71 configurations are suppressed as the quark mass falls . in fig .
[ dyn0.03q ] , [ dyn0.05q ] and [ dyn0.10q ] , the four correlators from @xmath64 configurations are compared with their counterparts from zero @xmath71 configurations for the three quark masses .
in contrast to the quenched data set , the effects of topology are small .
( note that not even the relative sizes of the correlators from different values of @xmath76 are the same for all three masses . )
the only place where topology appears to play a big role is in the @xmath43 correlator at the smallest quark mass .
even there , the large-@xmath71 correlator remains positive over the entire measured @xmath55 range .
based on our quenched results , we might have imagined a scenario in which the @xmath77 @xmath43 correlator was negative , but the correlator summed over all @xmath71 was positive , as required by unitarity .
that does not seem to be what is going on .
of course , the physics which is responsible for the probability of encountering a particular value of @xmath71 in an ensemble is almost certainly different in quenched and full qcd , so maybe it is not surprising that the correlation functions are different .
we studied the point to point light baryon correlators for four currents using quenched and two flavor dynamical overlap simulations . as far as we know , this is the first test of point to point correlators for light baryons in full qcd . in these four currents ,
there are three diquark structures : scalar ( @xmath27 and @xmath28 ) , pseudoscalar ( @xmath43 ) and axial vector ( @xmath59 ) .
these correlators are rather different in quenched approximation and in full qcd .
the quenched @xmath27 correlator is more attractive than in full qcd , and the quenched @xmath43 correlator is negative at small quark mass .
these features are correlated with the presence of topological zero modes in the quenched data set .
we have not attempted to make a direct comparison with models .
we are aware that in the instanton liquid model , baryon channels with different diquark quantum numbers show a strong response to the presence of topological zero modes @xcite
. it might be , that quenched results could be a diagnostic for models even though they are not reliable indicators of the behavior of correlators in full qcd .
the quark masses we are simulating are interesting ones because they go from a region where qualitatively speaking chiral symmetry effects are not so important ( @xmath78100 mev ) , to a region where they are probably important ( 30 - 40 mev ) .
as our quark mass falls , the dynamical results show apparent changes in each correlator ( see fig .
[ dyn_4j ] ) . in this quark mass range
, the scalar diquark created by @xmath0 seems to play a greater role in baryon correlators , than the others .
the @xmath27 correlator is the most attractive one , and the attraction in this channel rises strongly as the quark mass falls .
an attraction appears in the correlator of @xmath43 at the lowest quark mass in the dynamical simulation . in the light - light - heavy system
@xcite , the pseudoscalar diquark was shown to be heavier than the axial vector and scalar diquarks . here
the attraction in the @xmath43 correlator is slightly stronger than in the @xmath28 and @xmath59 correlators .
diquark effects in light baryons are apparently different from those in heavy baryons .
the big difference between the @xmath28 and @xmath27 correlators is also a strong statement that diquark correlations are not a dominant factor in light baryons .
both correlators contain scalar diquarks .
the difference between these correlators is how the diquark couples to the third quark , and this coupling apparently matters : we are not seeing an isolated diquark . finally , we have to conclude by again listing the shortcomings in this project : even our lightest quark masses are quite a bit heavier than the physical up and down quark masses .
simulations at smaller quark mass will require larger volume to prevent an effective restoration of chiral symmetry as the quark mass is reduced .
given the lattice artifacts we have seen in the quenched part of our study , we recommend against using mixed action or partially quenched simulations for these studies .
this work was supported by the us department of energy , french anr under grant anr-05-cigc-001 - 09 and by the deutsche forschungsgemeinschaft in the sfb / tr 09 .
we thank the desy computer center in zeuthen for essential computer resources and support .
this work was in part based on the milc collaboration s public lattice gauge theory code @xcite .
m. anselmino , e. predazzi , s. ekelin , s. fredriksson and d. b. lichtenberg , rev .
phys . * 65 * , 1199 ( 1993 ) .
r. babich , n. garron , c. hoelbling , j. howard , l. lellouch and c. rebbi , phys .
d * 76 * , 074021 ( 2007 ) [ arxiv : hep - lat/0701023 ] .
r. babich , n. garron , c. hoelbling , j. howard , l. lellouch and c. rebbi , pos * lat2006 * , 091 ( 2006 ) [ arxiv : hep - lat/0610079 ] . c. alexandrou , ph .
de forcrand and b. lucini , phys .
lett . *
97 * , 222002 ( 2006 ) [ arxiv : hep - lat/0609004 ] . c. alexandrou , p. de forcrand and b. lucini , pos * lat2005 * , 053 ( 2006 ) [ arxiv : hep - lat/0509113 ] .
k. orginos , pos * lat2005 * , 054 ( 2006 ) [ arxiv : hep - lat/0510082 ] .
z. fodor , c. hoelbling , m. mechtel and k. szabo , pos * lat2005 * , 310 ( 2006 ) [ arxiv : hep - lat/0511032 ] . m. hess , f. karsch , e. laermann and i. wetzorke , phys .
d * 58 * , 111502 ( 1998 ) [ arxiv : hep - lat/9804023 ] .
e. v. shuryak , rev .
phys . * 65 * , 1 ( 1993 ) .
r. l. jaffe , phys .
* 409 * , 1 ( 2005 ) [ nucl .
suppl . * 142 * , 343 ( 2005 ) ] [ arxiv : hep - ph/0409065 ] .
t. degrand , phys . rev .
d * 64 * , 094508 ( 2001 ) [ arxiv : hep - lat/0106001 ] .
a. hasenfratz , r. hoffmann and s. schaefer , jhep * 0705 * , 029 ( 2007 ) [ arxiv : hep - lat/0702028 ] .
see , for example , s. durr , z. fodor , c. hoelbling and t. kurth , jhep * 0704 * , 055 ( 2007 ) [ arxiv : hep - lat/0612021 ] .
w. a. bardeen , a. duncan , e. eichten , n. isgur and h. thacker , phys .
d * 65 * , 014509 ( 2001 ) [ arxiv : hep - lat/0106008 ]
. s. prelovsek , phys .
d * 73 * , 014506 ( 2006 ) [ arxiv : hep - lat/0510080 ] .
s. durr _ et al .
_ , arxiv:0710.4769 [ hep - lat ] .
c. aubin , j. laiho and r. s. van de water , arxiv:0710.0804 [ hep - lat ] .
see r. g. edwards , u. m. heller and r. narayanan , phys .
d * 59 * , 094510 ( 1999 ) [ arxiv : hep - lat/9811030 ] .
m. c. chu , j. m. grandy , s. huang and j. w. negele , phys .
d * 48 * , 3340 ( 1993 ) [ arxiv : hep - lat/9306002 ]
. t. schafer , e. v. shuryak and j. j. m. verbaarschot , nucl .
b * 412 * , 143 ( 1994 ) [ arxiv : hep - ph/9306220 ] .
s. j. hands , p. w. stephenson and a. mckerrell [ ukqcd collaboration ] , phys .
d * 51 * , 6394 ( 1995 ) [ arxiv : hep - lat/9412065 ] .
t. schafer and e. v. shuryak , phys .
d * 54 * , 1099 ( 1996 ) [ arxiv : hep - ph/9512384 ] .
t. schafer , e. v. shuryak and j. j. m. verbaarschot , phys .
d * 51 * , 1267 ( 1995 ) [ arxiv : hep - ph/9406210 ] .
m. cristoforetti , p. faccioli , g. ripka and m. traini , phys .
d * 71 * , 114010 ( 2005 ) [ arxiv : hep - ph/0410304 ] .
available from http://www.physics.utah.edu/ @xmath79 detar / milc / index.html . | we study the role of diquarks in light baryons through point to point baryon correlators .
we contrast results from quenched simulations with ones with two flavors of dynamical overlap fermions .
the scalar , pseudoscalar and axial vector diquarks are combined with light quarks to form color singlets . the quenched simulation shows large zero mode effects in correlators containing the scalar and pseudoscalar diquark .
the two scalar diquarks created by @xmath0 and @xmath1 lead to different behavior in baryon correlators , showing that the interaction of diquarks with the third light quark matters : we do not see an isolated diquark . in our quark mass range , the scalar diquark created by @xmath0 seems to play a greater role than the others . |
compressed sensing @xcite has generated enormous interest and research activities in mathematics , statistics , signal processing , imaging and information sciences , among numerous other areas .
one of the basic problems is to reconstruct a sparse signal under a few linear measurements ( linear constraints ) far less than the dimension of the ambient space of the signal . consider a sparse signal @xmath13 , an @xmath14 sensing matrix a and an observation @xmath15 , @xmath16 , such that : @xmath17 where @xmath18 is an @xmath19-dimensional observation error .
if @xmath20 is sparse enough , it can be reconstructed exactly in the noise - free case and in stable manner in the noisy case provided that the sensing matrix @xmath5 satisfies certain incoherence or the restricted isometry property ( @xmath21 ) @xcite .
the direct approach is @xmath1 optimization , including constrained formulation : @xmath22 and the unconstrained @xmath1 regularized optimization : @xmath23 with positive regularization parameter @xmath24 . since minimizing @xmath1 norm
is np - hard @xcite , many viable alternatives are available . greedy methods ( matching pursuit @xcite , othogonal matching pursuits ( omp ) @xcite , and regularized omp ( romp ) @xcite ) work well if the dimension @xmath19 is not too large . for the unconstrained problem ( [ eq : l0 uncons ] ) , the penalty decomposition method @xcite replaces the term @xmath25 by @xmath26 , and minimizes over @xmath27 for a diverging sequence @xmath28 .
the variable @xmath29 allows the iterative hard thresholding procedure .
the relaxation approach is to replace @xmath1 norm @xmath30 by a continuous sparsity promoting penalty functions @xmath31 .
convex relaxation uniquely selects @xmath32 as the @xmath0 norm @xmath33 .
the resulting problems are known as basis pursuit ( lasso in the over - determined regime @xcite ) .
the @xmath0 algorithms include @xmath0-magic @xcite , bregman and split bregman methods @xcite and yall1 @xcite . theoretically , cands and tao introduced rip condition and used it to establish the equivalent and unique global solution to @xmath1 minimization via @xmath0 relaxation among other stable recovery results @xcite .
there are also many choices of @xmath32 for non - convex relaxation .
one is the @xmath3 norm ( @xmath34 ) with @xmath1 equivalence under rip @xcite .
the @xmath35 norm is representative of this class of functions , with the reweighted least squares and half - thresholding algorithms for computation @xcite . near the rip regime
, @xmath35 penalty tends to have higher success rate of sparse reconstruction than @xmath0 .
however , it is not as good as @xmath0 if the sensing matrix is far away from rip @xcite as we shall see later as well . in the highly non - rip ( coherent ) regime
, it is recently found that the difference of @xmath0 and @xmath36 norm minimization gives the best sparse recovery results @xcite .
it is therefore of both theoretical and practical interest to find a non - convex penalty that is consistently better than @xmath0 and always ranks among the top in sparse recovery whether the sensing matrix satisfies rip or not . in the statistics literature of variable selection ,
fan and li @xcite advocated for classes of penalty functions with three desired properties : * unbiasedness , sparsity * and * continuity*. to help identify such a penalty function denoted by @xmath37 , fan and lv @xcite proposed the following condition for characterizing unbiasedness and sparsity promoting properties . the penalty function @xmath38 satisfies : 1 .
@xmath39 is increasing and concave in @xmath40 ; 2 .
@xmath41 is continuous with @xmath42 ; 3 .
if @xmath39 depends on a positive parameter @xmath24 , then @xmath43 is increasing in @xmath44 and @xmath45 is independent of @xmath24 .
it follows that @xmath41 is positive and decreasing , and @xmath45 is the upper bound of @xmath41 .
it is shown in @xcite that penalties satisfying condition 1 and @xmath46 enjoy both unbiasedness and sparsity .
though continuity does not generally hold for this class of penalty functions , a special one parameter family of functions , the so called * transformed @xmath0 functions ( tl1 ) @xmath47 * , where @xmath48 with @xmath49 , satisfies all three desired properties @xcite .
we shall study the minimization of tl1 functions for cs problems , in terms of theory , algorithms and computation .
we proposed the algorithms of tl1 via dc approximation @xcite and implemented numerical tests based on two classes of coherent random sensing matrices .
same as @xmath10 regularization @xcite , there also exists thresholding algorithm for tl1 , which is studied in the companion paper @xcite .
the rest of the paper is organized as follows . in section 2
, we study the properties of tl1 penalty and its regularization models .
one rip condition is given for the consistency of tl1 constrained model with original @xmath1 model .
we also prove that the local minimizers of the tl1 constrained model extract independent columns from the sensing matrix @xmath5 , as well as the local minimizers of the unconstrained model . in section 3 ,
we present two dc algorithms for tl1 optimization ( dcatl1 ) . in section 4 , we compare the performance of dcatl1 with some state - of - the - art methods using two classes of matrices : the gaussian and the oversampled discrete cosine transform ( dct ) .
numerical experiments indicate that dcatl1 is robust and consistently top ranked while maintaining high sparse recovery rates across all sensing matrices .
concluding remarks are in section 5 .
[ cols= " < , > " , ]
we have studied compressed sensing problem with the transformed @xmath0 penalty function for both the unconstrained and constrained models .
we presented a theory on the uniqueness and @xmath1 equivalence of the global minimizer of the unconstrained model under rip and analyzed properties of local minimizers .
we showed two dc algorithms along with a convergence theory . in numerical experiments
, dcatl1 is on par with the best method reweighted @xmath35 ( @xmath50 ) in the unconstrained ( constrained ) model , using incoherent gaussian matrices @xmath5 . for highly coherent over - sampled dct matrices ,
dcatl1 is comparable to the best method dca @xmath51 algorithm . for random matrices of varying degree of coherence
, we tested gaussian and over - sampled dct sensing matrices .
the dcatl1 algorithm is the most robust for constrained and unconstrained models alike . in future work , we plan to develop tl1 algorithms for imaging processing applications such as deconvolution and deblurring .
the authors would like to thank professor wenjiang fu for suggesting reference @xcite and a helpful discussion .
we also wish to thank the reviewers for their constructive comments .
the proof generally follows the lines of arguments in @xcite and @xcite , while using special properties of the penalty function @xmath52 .
+ for simplicity , we denote @xmath53 by @xmath54 and @xmath55 by @xmath56 .
+ define the function : @xmath57 it is continuous and increasing in the parameter @xmath58 . note that at @xmath59 , @xmath60 , and as @xmath61 , @xmath62 by ( [ rip ] ) .
there exists a constant @xmath7 , such that @xmath63 .
the number @xmath7 depends on the rip of matrix @xmath5 only , and so it is independent of the scalar @xmath64 . + for @xmath65 : @xmath66 let @xmath67 , and we want to prove that the vector @xmath68 .
it is clear that , @xmath69 , since @xmath70 is the support set of @xmath56 . by the triangular inequality of @xmath52 , we have : @xmath71 then @xmath72 it follows that : @xmath73 now let us arrange the components at @xmath74 in the order of decreasing magnitude of @xmath75 and partition into @xmath76 parts : @xmath77 , where each @xmath78 has @xmath79 elements ( except possibly @xmath80 with less ) .
also denote @xmath81 and @xmath82 .
since @xmath83 , it follows that @xmath84 now we estimate the @xmath36 norm of @xmath88 from above in terms of @xmath85 .
it follows from @xmath54 being the minimizer of the problem ( [ eq : spar revised ] ) and the definition of @xmath89 ( [ para : scaled pa ] ) that @xmath90 by ( [ rip a ] ) , the factor @xmath100 is strictly positive , hence @xmath101 , and @xmath102 . also by inequality ( [ ineq : ttc ] ) , @xmath103 .
we have proved that @xmath104 .
the equivalence of ( [ eq : spar revised ] ) and ( [ eq : l0 revised ] ) holds .
d. needell and r. vershynin , _ signal recovery from incomplete and inaccurate measurements via regularized orthogonal matching pursuit , _ ieee journal of selected topics in signal processing , 4(2):310 - 316 , 2010 .
s. zhang and j. xin , _ minimization of transformed @xmath108 penalty : closed form representation and iterative thresholding algorithms _ , arxiv preprint , arxiv:1412.5240 ( 2014 ) , comm .
math sci , to appear . | we study the minimization problem of a non - convex sparsity promoting penalty function , the transformed @xmath0 ( tl1 ) , and its application in compressed sensing ( cs ) .
the tl1 penalty interpolates @xmath1 and @xmath0 norms through a nonnegative parameter @xmath2 , similar to @xmath3 with @xmath4 $ ] .
tl1 is known in the statistics literature to enjoy three desired properties : unbiasedness , sparsity and lipschitz continuity .
we first consider the constrained minimization problem and prove the uniqueness of global minimizer and its equivalence to @xmath1 norm minimization if the sensing matrix @xmath5 satisfies a restricted isometry property ( rip ) and if @xmath6 , where @xmath7 depends only on @xmath5 . though result contains the well - known equivalence of @xmath0 norm and @xmath1 norm , in the limit @xmath8 , the main difficulty is in treating the lack of scaling property of the tl1 penalty function . for a general sensing matrix @xmath5
, we show that the support set of a local minimizer corresponds to linearly independent columns of @xmath5 , and recall sufficient conditions for a critical point to be a local minimum .
next , we present difference of convex algorithms for tl1 ( dcatl1 ) in computing tl1-regularized constrained and unconstrained problems in cs .
the dcatl1 algorithm involves outer and inner loops of iterations , one time matrix inversion , repeated shrinkage operations and matrix - vector multiplications . for the unconstrained problem , we prove convergence of dcalt1 to a stationary point satisfying the first order optimality condition . finally in numerical experiments ,
we identify the optimal value @xmath9 , and compare dcatl1 with other cs algorithms on two classes of sensing matrices : gaussian random matrices and over - sampled discrete cosine transform matrices ( odct ) . among existing algorithms ,
the iterated reweighted least squares method based on @xmath10 norm is the best in sparse recovery for gaussian matrices , and the dca algorithm based on @xmath11 penalty is the best for odct matrices .
we find that for both classes of sensing matrices , the performance of dcatl1 algorithm ( initiated with @xmath12 minimization ) always ranks near the top ( if not the top ) , and is the _ most robust choice _ insensitive to rip ( incoherence ) of the underlying cs problems .
* keywords : * transformed @xmath0 penalty , sparse signal recovery theory , difference of convex function algorithm , convergence analysis , coherent random matrices , compressed sensing , robust recovery 90c26 , 65k10 , 90c90 |
in the last decade interesting correspondences were discovered relating quiver gauge theories , lattice polygons and calabi - yau singularities .
the motivation and evolution of these ideas in physics are well - documented in many articles ; e.g. @xcite . in the present paper
we want to put some aspects of these correspondences from physics alongside the hypergeometric systems in two variables of gelfand , kapranov , zelevinsky @xcite and the dessins denfants of grothendieck et al .
this reveals intriguing connections between these fields .
the beautiful insight of gelfand , kapranov and zelevinsky was that hypergeometric structures greatly simplify if one introduces extra variables and balances this with an appropriate torus action @xcite . in order to profit from the simplication they developed tools like the _ secondary fan , secondary polytope _ and _
principal @xmath0-determinant_. this paper demonstrates that these are also very practical tools for studying quivers and dimer models . a similar beautiful insight of simplification by going to higher dimensions appeared in de bruijn s construction @xcite of penrose tilings and developed into the well - known projection method in the theory of quasi - crystals ; e.g. @xcite .
we apply the same method to construct periodic rhombus tilings of the plane , which the physicists call _ brane tilings _ and _ dimer models_. from a geometric perspective this paper deals with embeddings of quivers into compact oriented surfaces without boundary .
more specifically , the initial combinatorial data for a quiver @xmath1 are two finite sets @xmath2 ( arrows ) and @xmath3 ( nodes ) and two maps @xmath4 ( source and target ) .
embedding @xmath1 into a compact oriented surface without boundary @xmath5 means that @xmath3 becomes a subset of @xmath5 and an arrow @xmath6 becomes a path @xmath7 in @xmath5 from the point @xmath8 to the point @xmath9 .
it is required that the boundary of every _ face of _
@xmath10 i.e. connected component of @xmath11 is formed by a sequence of paths @xmath12 with @xmath13 for @xmath14 ; @xmath15 .
it is evident from this requirement that the face boundaries receive an orientation from @xmath1 .
when this is combined with the orientation on @xmath5 faces lie either on the positive or on the negative side of their boundary . in pictures we mark faces which lie on the positive ( respectively negative ) side of their boundary with @xmath16 ( resp .
@xmath17 ) .
it is obvious that adjacent faces get different colors .
the dual graph of @xmath1 w.r.t .
@xmath5 is thus a bi - partite graph @xmath18 , embedded in @xmath5 : the nodes of @xmath18 correspond with the faces of @xmath10 ; the edges of @xmath18 connect nodes coming from adjacent faces and correspond bijectively with the edges of @xmath1 .
a pair @xmath10 consisting of a compact oriented surface without boundary and an oriented graph embedded in it is called a _
dessin denfants _ or just _ dessin _ , in one of the various equivalent definitions of `` dessin ( denfants ) '' ; see @xcite .
other definitions refer to @xmath19 as dessin denfants . in case @xmath5
has genus @xmath20 one often calls @xmath18 and its lifting to the plane ( i.e. the universal covering of the torus @xmath5 ) a _ dimer model _ or _ brane tiling _ ; see for instance @xcite .
yet another presentation of the same structure gives the dessin as the triangulation of @xmath5 with vertex set the union of the nodes of @xmath1 and @xmath18 ; the triangles are given by the triplets of vertices consisting of two nodes of @xmath18 connected by an edge @xmath21 of @xmath18 and one node @xmath22 of @xmath1 incident to the edge @xmath23 of @xmath1 which is dual to @xmath21 .
figure [ fig : p2combi ] shows these three get - ups of the dessin denfants of the @xmath24-quiver ( case @xmath25 in figure [ fig : models ] ) ( 350,100)(0,-5 ) ( -15,10 ) ( 150,100)(0,15 ) ( 120,10 ) ( 150,100)(0,15 ) ( 255,10 ) ( 150,100)(0,15 ) from an algebraic perspective this paper deals with _ superpotentials for quivers_. there is a simple equivalence between the geometric and algebraic perspectives : the superpotential is a convenient way of writing the list of oriented boundaries @xmath12 of the faces of @xmath10 with @xmath26-signs indicating how the orientation matches with that of @xmath5 .
it can also be read as the instruction for building @xmath5 by glueing polygons . for the dessin denfants in figure [ fig : p2combi ] the superpotential ( for some numbering of the arrows of @xmath1 )
is @xmath27 another algebraic perspective , equivalent to the previous one , is given by what we want to call _ the bi - adjacency matrix of the dessin denfants @xmath10 with weight @xmath28_. it is defined as follows .
weight _ is a function @xmath29 .
the dessins in the present paper have as many black faces as white faces and every edge @xmath6 lies in the boundary of a unique black face , denoted @xmath30 , and a unique white face , denoted @xmath31 .
the bi - adjacency matrix of @xmath10 with weight @xmath28 is a square matrix @xmath32 with rows corresponding with the black faces , columns corresponding with the white faces and entries in the polynomial ring @xmath33 $ ] which has one variable for every node @xmath22 of the quiver @xmath1 : _ the @xmath34-entry of @xmath32 is _ @xmath35 the bi - adjacency matrix is in fact the _ kasteleyn matrix _ of a twist of the dimer model @xmath19 ; see section [ sec : kasteleyn superpot ] . for the dessin in figure [ fig : p2combi ]
the bi - adjacency matrix ( for some numbering of the nodes , edges and faces ) is @xmath36\,.\ ] ] in section [ sec : quiver ] we describe how the quivers for which we can solve the embedding problem , are associated with certain rank @xmath37 subgroups @xmath38 of @xmath39 ; here @xmath40 is the number of nodes of the quiver .
such subgroups @xmath41 are the foundations for the theory of gelfand , kapranov and zelevinsky . in section
[ sec : lsecondaries ] we describe the _ secondary fan _ and _ the secondary polygon _ associated with @xmath41 .
these important structures in gkz - theory surprisingly turn out to be also quite relevant for the dimer models .
we show for instance in theorem [ thm : kasteleyn and secondary polygon ] that the secondary polygon is the _ newton polygon _ of the determinant of the bi - adjacency matrix with non - zero weights . in section [ sec : a determinant ] we report the observation , based on examples , that for the _ critical weight _
@xmath42 the determinant of the bi - adjacency matrix @xmath43 , after a simple transformation , becomes equal to the _ principal @xmath0-determinant _ of gelfand , kapranov , zelevinsky @xcite . in sections
[ sec : hyper ] and [ sec : solutions ] we recall some topics in the theory of gkz - hypergeometric systems of differential equations . section [ sec : dessin ] presents the algorithm to solve the quiver embedding problem .
it is a combination of the _ fast inverse algorithm _ of @xcite and the _ untwisting procedure _ of @xcite . in the cited papers ,
however , methods are mainly presented via visual inspection of pictures in some concrete examples .
so an extrapolation to general situations was needed .
the first impression that the fast inverse algorithm of @xcite is more or less de bruijn s construction @xcite of penrose tilings did not quite yield the sought after embedding of the quiver ; the untwisting procedure of @xcite is also needed .
the algorithm became a smoothly operating algebraic - combinatorial tool by consistently working from the philosophy that things look simpler from a higher dimensional viewpoint ( with group action ) .
for doing the computer experiments behind this paper i implemented the algorithm in matlab .
the reader can find in section [ sec : dessin ] sufficient details for making a computer version of the algorithm .
sections [ sec : perfect sec fan][sec : a determinant ] demonstrate that consistently working from the higher dimensional viewpoint leads to new insights in the dimer technology tools _ perfect matchings , kasteleyn matrix and its determinant _ and shows their close relation with the gkz tools _ secondary polytope and principal @xmath0-determinant_. besides problems such as proving that the algorithm of section [ sec : dessin ] yields at least one superpotential for every quiver which satisfies the conditions in theorem [ quivergrassmann ] , or proving conjecture [ conj : critical weight ] our work raises some interesting questions like : * q1 . *
the calabi - yau singularities in the background of this work are constructed from toric diagrams , which here are interpreted as secondary polygons . in @xcite
6 the singularity is obtained from the toric diagram by a symplectic quotient construction . in @xcite
2 the singularity is given as a toric variety constructed from a fan with one maximal cone , namely the cone over the toric diagram ( polygon ) .
for this the polygon is put into the plane of points with first coordinate @xmath20 in @xmath44 . in the work of gelfand , kapranov and
zelevinsky the secondary polygon is put into a @xmath37-dimensional plane in @xmath45 , which does not pass through @xmath46 ; here @xmath40 is the number of nodes of the quiver .
one can perform the standard toric variety construction with the @xmath47-dimensional cone over the secondary polygon in @xmath45 .
_ does this toric variety give the appriopriate view in the gkz philosophy ( higher dimension compensated by group action ) on the singular calabi - yau @xmath47-space ? _ since this toric variety is a natural domain for gkz hypergeometric functions ( see [ toric sec fan ] ) one may wonder : _ do hypergeometric functions provide new useful tools for investigating the geometry of @xmath47-dimensional calabi - yau singularities ? _ possibly positive indications are the fact that the dimension of the solution space of the gkz system of hypergeometric differential equations equals the size of the bi - adjacency matrix @xmath43 ( see theorem [ thm : solution dimension ] and corollary [ black vola ] ) and the observation ( conjecture [ conj : critical weight ] ) that the determinant of @xmath43 equals , up to a simple transformation , the principal @xmath0-determinant , which describes the singularities of the gkz system .
* q2 . * the surface @xmath5 with embedded in it the pair of dual graphs @xmath1 and
@xmath18 has been constructed in a purely combinatorial topological way from the superpotential .
it is a general fact ( see @xcite 1.2 ) that there is then a continuous map @xmath48 from @xmath5 to the @xmath37-sphere @xmath49 which is a ramified covering with exactly three ramification points such that the set of vertices of @xmath1 , the set of black vertices of @xmath18 and the set of white vertices of @xmath18 are the fibers over the three ramification points .
a highlight in the theory of dessins denfants is _ belyi s theorem _ ( see @xcite theorem 2.1.1 ) .
it states that in the above situation the surface @xmath5 admits a model @xmath50 over a number field @xmath51 such that @xmath48 becomes a morphism @xmath52 of algebraic curves over @xmath51 which is unramified outside @xmath53 .
the labeling of the ramification points can be taken such that @xmath54 , @xmath55 and @xmath56 are the sets of white and black vertices of @xmath18 and the vertices of @xmath1 , respectively .
the bipartite graph @xmath18 in @xmath5 is then the inverse image of the negative real axis @xmath57 $ ] in the riemann sphere @xmath58 and the quiver @xmath1 is the inverse image of the positively oriented unit circle @xmath59 .
it is usually difficult to find explicit algebraic equations for @xmath50 and the _ belyi function _ @xmath48 .
+ on the other hand , the authors of @xcite write in footnote @xmath60 : _ .... we have produced a dimer model that is defined on its own spectral curve @xmath61 . _
unfortunately their arguments are not ( yet ) sufficiently refined to yield the weight @xmath28 that is to be used in this equation .
+ it seems an interesting challenge to tackle the two problems simultaneously and look for a weight @xmath28 and a belyi function @xmath48 on the algebraic curve with equation @xmath61 that realize @xmath62 as described above .
* .. the most basic object*[introl ] in this paper is a rank @xmath37 subgroup @xmath41 which is not contained in any of the standard coordinate hyperplanes of @xmath39 and is perpendicular to the vector @xmath63 , with all coordinates @xmath20 .
so , the elements of @xmath38 are vectors @xmath64 with @xmath65 and for every @xmath66 there is an @xmath67 such that @xmath68 . *
.. notation.*[nota : ej ] throughout this paper @xmath69 is the standard basis of @xmath39 and @xmath70 denotes the matrix @xmath71 . *
.. definition / construction.*[plucker coordinates ] taking the second exterior powers one finds an inclusion @xmath72 .
the group @xmath73 is a free @xmath74-module of rank @xmath20 .
after fixing the orientation on @xmath38 , i.e. choosing one of the two possible isomorphisms @xmath75 , one gets an inclusion @xmath76 .
the coordinates of @xmath77 with respect to the standard basis @xmath78 of @xmath79 are called the _ plcker coordinates of @xmath41_. dual to the above inclusion one finds the map @xmath80 , i.e. an anti - symmetric bilinear form on @xmath39 . in view of its relation with the plcker coordinates we call this the _ plcker form of @xmath38_. to make this look more explicit we take a basis for @xmath38 compatible with the chosen orientation .
one can represent the two basis vectors as the rows of a @xmath81-matrix @xmath82 .
let @xmath83 be the columns of this matrix .
then the plcker coordinates are @xmath84 and the ( anti - symmetric ) matrix for the plcker form is @xmath85 .
the plcker coordinates and the plcker form do , of course , not change if one takes another basis for @xmath38 with the same orientation . note that the matrix for the plcker form has rank @xmath37 : @xmath86 * .. definition.*[quiver basics ] a _ quiver _
@xmath1 is a finite directed graph .
it can be given combinatorially by two finite sets @xmath2 and @xmath3 and two maps @xmath4 ; in short hand notation @xmath87 .
each element of @xmath3 is a node of the graph and an element @xmath6 is an arrow from node @xmath8 to node @xmath9 .
the _ adjacency matrix _ of the quiver is the matrix @xmath88 with entry @xmath89 equal to the number of arrows from node @xmath90 to node @xmath91 . if the quiver has no directed loops of length @xmath92 ( i.e. @xmath93 for all @xmath94 ) , it is faithfully described by the _ anti - symmetrized adjacency matrix _ @xmath95 .
if the @xmath96-entry of @xmath97 is positive it equals the number of arrows from node @xmath90 to node @xmath91 .
if it is negative it is the number of arrows from @xmath91 to @xmath90 . * .. definition.*[pluckerquiver ] we call the quiver without directed loops of length @xmath92 with anti - symmetrized adjacency matrix equal to the matrix of the plcker form of @xmath38 the _ plcker quiver of @xmath38_. * .. examples * of groups @xmath41 and their plcker quivers are shown in figures [ fig : models ] and [ fig : dp3 models ] . * .. * [ kirchhof ] some easily visible geometric properties of a quiver without directed loops of length @xmath92 are equivalent to easily seen algebraic properties of its anti - symmetrized adjacency matrix @xmath97 .
for instance , the graph has no isolated nodes if and only if no column of @xmath97 is zero .
also , in every node of the quiver the number of incoming arrows equals the number of outgoing arrows precisely if in each row of @xmath97 the sum of the entries is @xmath98 . for plcker quivers
these properties correspond to the conditions that @xmath38 is not contained in any of the standard coordinate hyperplanes of @xmath39 and is perpendicular to the vector @xmath63 .
one property of a plcker quiver which one can not easily read off from the graph is @xmath99 . in the remainder of this section
we show that these properties characterize plcker quivers of rank @xmath37 subgroups @xmath41 as in [ introl ] . * .. proposition.*[prop : rank 2 anti - sym ] _
an @xmath100-matrix @xmath101 with entries in @xmath74 is anti - symmetric and @xmath102 if and only if there is a @xmath81-matrix @xmath82 with entries in @xmath74 such that @xmath103 and @xmath104 .
_ * proof . *
the `` if''-statement is trivial .
so let us consider the `` only if '' and assume that @xmath101 is anti - symmetric and @xmath102 .
let @xmath105 denote the greatest common divisor of the entries of @xmath101 and @xmath106 .
choose an @xmath107-matrix @xmath108 whose columns form a @xmath74-basis for the column space of @xmath109 .
the equality of column spaces of @xmath109 and @xmath108 means that there are an @xmath107-matrix @xmath2 and a @xmath81-matrix @xmath110 , both with entries in @xmath74 and of rank @xmath37 , such that @xmath111 and @xmath112 .
then @xmath113 so @xmath114 is an anti - symmetric @xmath115-matrix ; say @xmath116 with @xmath117 .
in fact @xmath118 since it divides all entries of @xmath109 .
if @xmath119 we replace @xmath110 by @xmath120 .
we then always have @xmath121 .
let @xmath122 be any @xmath115-matrix with entries in @xmath74 and @xmath123 .
let @xmath124 .
then @xmath125 as wanted .
@xmath126 * .. * [ quiver to l ] multiplying in proposition [ prop : rank 2 anti - sym ] the matrix @xmath82 from the left by a matrix from @xmath127 does not change the matrix @xmath128 .
so it is more natural to interpret @xmath101 as the matrix of the plcker form of the @xmath74-row space of @xmath82 .
note that it follows from the proof of proposition [ prop : rank 2 anti - sym ] that this space is uniquely determined if the greatest common divisor of the entries of @xmath101 is @xmath20 .
if on the other hand the greatest common divisor of the entries of @xmath101 is @xmath129 this space depends on the additional choice of a @xmath115-matrix with entries in @xmath74 and determinant @xmath105 .
let us summarize the above discussion : * .. theorem.*[quivergrassmann ] _ for every @xmath41 as in [ introl ] one has its plcker quiver .
this quiver has no isolated nodes or directed loops of length @xmath92 and in every node the number of incoming arrows equals the number of outgoing arrows
. moreover the rank of its anti - symmetrized adjacency matrix is @xmath37 .
conversely , every quiver with these properties is the plcker quiver of some @xmath41 as in [ introl ] .
the correspondence between such quivers and such @xmath41 is one - to - one for those quivers for which the greatest common divisor of the entries of the anti - symmetrized adjacency matrix is @xmath20 and those @xmath41 for which @xmath130 has no torsion .
* .. * [ intro l secondary ] for @xmath41 as in [ introl ] we now define the _ secondary fan _ and the _ secondary polytope_. in [ a secondary ] we will compare this with the original definitions by gelfand , kapranov and zelevinsky . the term `` secondary '' refers to the fact that in their theory of hypergeometric systems another polytope appears first , which is therefore called the _
primary polytope_. nonetheless both the secondary fan and the secondary polytope can most conveniently and directly be described using the lattice @xmath38 . in case
the rank of @xmath38 is @xmath37 , the constructions become particularly simple . * .. definition.*[def : lsecondary fan ] let @xmath41 be as in [ introl ] .
let @xmath131 denote the real dual space of @xmath38 .
let @xmath69 be the standard basis of @xmath39 .
let @xmath132 be the image of @xmath133 under the map @xmath134 dual to the inclusion @xmath135 . here and
henceforth we identify @xmath45 with the real dual space of @xmath39 by means of the standard dot product . by definition ,
the _ secondary fan of @xmath38 _ is the following collection of cones in @xmath136 : the @xmath98-dimensional cone @xmath137 , the @xmath20-dimensional cones @xmath138 ( @xmath139 ) and the @xmath37-dimensional cones which are the closures of the connected components of @xmath140 .
* .. definition.*[def : lsecondary polytope ] with a @xmath37-dimensional cone @xmath101 in the secondary fan one associates the set @xmath141 of @xmath37-element subsets of @xmath142 and the vector @xmath143 the _ secondary polytope _
@xmath144 of @xmath41 is then defined as @xmath145 * .. remark.*[secondary interpretations ] in [ a secondary ] we give an interpretation of the above @xmath146 as a triangulation of the primary polytope . in equation ( [ eq : pclc ] ) we associate with @xmath146 a perfect matching in a bipartite graph . * .. example.*[secondary fan dp1 ] figure [ fig : secondaries ] shows the secondary fan with the lists @xmath146 , the secondary polytope @xmath147 with coordinates for the vertices and the primary polytope for @xmath148 .
this is case @xmath149 of figure [ fig : models ] ; see also example [ cubic roots ] and [ a secondary ] .
( 300,170)(-10,-50 ) ( 260,10 ) ( 100,80)(0,0 ) ( 0,0)(0,-1)40 ( 0,0)(1,0)40 ( 0,0)(1,2)40 ( 0,0)(-2,-1)80 ( -1,-51)@xmath150 ( 42,-2)@xmath151 ( 41,83)@xmath152 ( -89,-44)@xmath153 ( 15,-33)@xmath154 $ ] ( 25,30)@xmath155 $ ] ( -47,-40)@xmath156 $ ] ( -50,5)@xmath157 $ ] ( -10,70 ) ( 60,80)(-10,-10 ) ( 5,10)@xmath158 $ ] ( 35,10)@xmath159 $ ] ( 65,10)@xmath160 $ ] ( 95,10)@xmath161 $ ] ( 20,-10)(1,0)90 ( 20,-10 ) ( 50,-10 ) ( 80,-10 ) ( 110,-10 ) ( 40,0 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,-30 ) ( 30,-30 ) ( 60,30 ) ( 30,0 ) ( 0,0)(0,-1)30 ( 0,-30)(1,0)30 ( 30,-30)(1,2)30 ( 60,30)(-2,-1)60 ( -34,10)@xmath162 $ ] ( -34,-45)@xmath163 $ ] ( 30,-45)@xmath164 $ ] ( 30,35)@xmath165 $ ] * .. * [ secondary polytope from quiver ] the geometric formulation of the construction of the secondary fan and polytope is quite attractive . nonetheless in practical algorithms everything
can be most easily obtained from the matrix of the plcker form of @xmath41 .
indeed , a @xmath37-dimensional cone in the secondary fan is bounded by two half - lines @xmath138 and @xmath166 with the property that @xmath167 and @xmath168 for @xmath169 .
thus to find the secondary fan one just needs to find all pairs @xmath170 with these properties . for a cone @xmath101 bounded by @xmath138 and @xmath166
the list @xmath146 is then @xmath171 now consider two adjacent @xmath37-dimensional cones in the secondary fan , say @xmath101 and @xmath109 .
let @xmath172 be those vectors from the set @xmath173 that lie on the half - line @xmath174 .
without loss of generality we may assume @xmath101 lies to the right of @xmath175 and @xmath109 to the left .
then @xmath176 from this we see , using @xmath177 for all @xmath90 , that @xmath178 in other words , @xmath179 is the sum of rows @xmath180 of the matrix of the plcker form . * .. * [ explict secondary polygon ] to make this even more explicit and simple looking we take a basis for @xmath38 compatible with the chosen orientation .
one can represent the two basis vectors as the rows of a @xmath81-matrix @xmath82 .
let @xmath83 be the columns of this matrix .
the elements of @xmath38 should now be written as row vectors with two components and the embedding @xmath135 is given by @xmath181 . since @xmath182 , we can reformulate equation ( [ eq : secondary edge ] ) as @xmath183 in order to obtain the simplest formulation for the construction we order the vectors @xmath83 so that the points @xmath184 for @xmath169 lie ordered counterclockwise on the boundary of the polygon @xmath185 _ then the secondary polygon @xmath144 is obtained by first rotating @xmath186 clockwise over @xmath187 , next embedding it along with @xmath38 into @xmath39 and finally translating it over the vector @xmath188 , where @xmath101 is the cone in the secondary fan with left hand boundary @xmath189 . _
note that while the secondary polytope @xmath147 depends only on the embedding @xmath135 , the polygon @xmath186 usually changes when one puts another vector @xmath190 in first position by a cyclic permutation or when one multiplies the vectors @xmath83 by a matrix from @xmath127 .
* .. * [ polygon to l ] for a converse to the above construction we start from a convex polygon @xmath186 in @xmath191 with vertices in @xmath192 and non - empty interior .
let @xmath193 denote its boundary .
next we choose a collection of points @xmath194 in @xmath195 which includes all vertices of @xmath186 .
we number these points so that @xmath196 and @xmath197 are consecutive points in the counter - clockwise orientation of @xmath193 .
thinking of the elements of @xmath192 as row vectors we define column vectors @xmath198 and @xmath199 for @xmath200 .
finally we define @xmath38 to be the @xmath74-row space of the @xmath81-matrix @xmath82 with columns @xmath83 .
_ thus every convex polygon @xmath186 in @xmath191 with vertices in @xmath192 and non - empty interior can be viewed as the secondary polygon of some rank @xmath37 subgroup @xmath41 satisfying the conditions in [ introl ] . _ note however that in general there are several possible choices for the points @xmath194 in @xmath195 .
the minimal choice takes only the vertices of @xmath186 , while the maximal choice takes all points of @xmath195 .
* .. theorem.*[secondary area ] _ the ( euclidean ) area of the polygon @xmath186 in ( [ eq : easy secondary ] ) is _
@xmath201 _ for every @xmath37-dimensional cone @xmath101 in the secondary fan .
_ * proof . * after applying , if necessary , a cyclic permutation to @xmath83 we may assume , without loss of generality , that @xmath202 . from the triangulation @xmath186 by the diagonals between the vertex @xmath203 and the other vertices of
@xmath186 one sees that @xmath204 where @xmath205 is such that @xmath206 for @xmath207 and @xmath208 for @xmath209 .
the fact that @xmath210 , implies @xmath211 thus we see @xmath212 next note that for @xmath213 the inequality @xmath214 holds if and only if @xmath215 , if and only if @xmath216 . equality ( [ eq : aread ] ) now follows immediately .
@xmath126 * .. corollary.*[internal secondary ] _ the number of lattice points in the interior of the polygon @xmath186 in ( [ eq : easy secondary ] ) is _
@xmath217 * proof .
* this follows from theorem [ secondary area ] in combination with _
pick s formula _
( @xcite p.113 ) : @xmath218 @xmath126 ( 340,560)(-30,0)(0,480 ) ( 300,80)(-20,-10 ) ( 0,0 ) ( 100,80)(0,0 ) ( 0,0 ) ( 60,0 ) ( 30,45 ) ( 0,0)(1,0)57 ( 0,0)(1,0)47 ( 0,0)(1,0)37 ( 60,0)(-2,3)29 ( 60,0)(-2,3)24 ( 60,0)(-2,3)19 ( 30,45)(-2,-3)29 ( 30,45)(-2,-3)24 ( 30,45)(-2,-3)19 ( 70,0 ) ( 60,80)(-10,-10 ) ( 10,20)@xmath219 \left [ \begin{array}{rrr } 2&-1&-1\\ 1&1 & -2 \end{array}\right ] \end{array}$ ] ( 20,-20)model of @xmath24 ( 250,0 ) ( 100,80)(0,0 ) ( 0,0 ) ( 60,30 ) ( 30,60 ) ( 30,30 ) ( 0,0)(2,1)60 ( 60,30)(-1,1)30 ( 30,60)(-1,-2)30 ( 0,360 ) ( 300,80)(-10,-10 ) ( 0,10 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,60 ) ( 60,0 ) ( 60,60 ) ( -10,0)@xmath220 ( -10,60)@xmath20 ( 65,0)@xmath47 ( 65,60)@xmath37 ( 0,0)(0,1)57 ( 0,0)(0,1)53 ( 0,0)(1,1)57 ( 60,0)(-1,0)57 ( 60,0)(-1,0)53 ( 60,0)(-1,0)49 ( 0,60)(1,0)57 ( 0,60)(1,-1)57 ( 60,60)(0,-1)57 ( 60,60)(0,-1)53 ( 70,10 ) ( 60,80)(-10,-10 ) ( 20,20)@xmath221 \left[\begin{array}{rrrr } 0&1&1&-2\\ -1&0&2&-1 \end{array}\right]\end{array}$ ] ( 33,-20)model of @xmath222 ( 250,40 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,-30 ) ( 30,-30 ) ( 60,30 ) ( 30,0 ) ( 0,0)(0,-1)30 ( 0,-30)(1,0)30 ( 30,-30)(1,2)30 ( 60,30)(-2,-1)60 ( 0,240 ) ( 300,80)(-10,-10 ) ( 0,10 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,60 ) ( 60,0 ) ( 60,60 ) ( -10,0)@xmath20 ( -10,60)@xmath220 ( 65,0)@xmath47 ( 65,60)@xmath37 ( 0,60)(1,0)57 ( 0,60)(1,0)53 ( 0,60)(0,-1)57 ( 0,60)(0,-1)53 ( 0,0)(1,0)57 ( 0,0)(1,0)53 ( 60,0)(-1,1)57 ( 60,0)(-1,1)53 ( 60,0)(-1,1)49 ( 60,0)(-1,1)45 ( 60,60)(0,-1)57 ( 60,60)(0,-1)53 ( 70,10 ) ( 60,80)(-10,-10 ) ( 20,20)@xmath223 \left[\begin{array}{rrrrrr } 1&1&-1&-1\\ 0&0&2&-2 \end{array}\right]\end{array}$ ] ( 33,-20)model i of @xmath224 ( 250,10 ) ( 100,80)(0,0 ) ( 0,0 ) ( 30,0 ) ( 60,0 ) ( 30,60 ) ( 30,30 ) ( 0,0)(1,0)60 ( 60,0)(-1,2)30 ( 30,60)(-1,-2)30 ( 0,120 ) ( 300,80)(-10,-10 ) ( 0,10 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,60 ) ( 60,0 ) ( 60,60 ) ( -10,0)@xmath220 ( -10,60)@xmath20 ( 65,0)@xmath47 ( 65,60)@xmath37 ( 0,0)(0,1)57 ( 0,0)(0,1)53 ( 60,0)(-1,0)57 ( 60,0)(-1,0)53 ( 0,60)(1,0)57 ( 0,60)(1,0)53 ( 60,60)(0,-1)57 ( 60,60)(0,-1)53 ( 70,10 ) ( 60,80)(-10,-10 ) ( 20,20)@xmath225 \left[\begin{array}{rrrrrr } 1&1&-1&-1\\ -1&1&1&-1 \end{array}\right]\end{array}$ ] ( 33,-20)model ii of @xmath224 ( 250,40 ) ( 100,80)(0,0 ) ( 0,0 ) ( 30,-30 ) ( 60,0 ) ( 30,30 ) ( 30,0 ) ( 0,0)(1,-1)30 ( 30,-30)(1,1)30 ( 60,0)(-1,1)30 ( 30,30)(-1,-1)30 ( 0,0 ) ( 300,80)(-10,-10 ) ( 0,30 ) ( 100,80)(0,0 ) ( 0,0 ) ( 30,0 ) ( 30,-30 ) ( 60,0 ) ( 0,30 ) ( -10,27)@xmath20 ( -10,-3)@xmath37 ( 28,-40)@xmath220 ( 65,-3)@xmath47 ( 28,5)@xmath226 ( 0,0)(1,-1)27 ( 0,0)(1,-1)23 ( 30,-30)(-1,2)27 ( 30,-30)(0,1)27 ( 30,-30)(0,1)23 ( 30,-30)(0,1)19 ( 30,0)(-1,1)27 ( 30,0)(-1,0)27 ( 30,0)(1,0)27 ( 0,30)(2,-1)57 ( 0,30)(0,-1)27 ( 60,0)(-1,-1)27 ( 60,0)(-1,-1)23 ( 70,10 ) ( 60,80)(-10,-10 ) ( 0,20)@xmath227 \left[\begin{array}{rrrrrr } 0&1&1&-1&-1\\ -1&0&0&2&-1 \end{array}\right]\end{array}$ ] ( 30,-20)model i of @xmath228 ( 250,30 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,-30 ) ( 30,-30 ) ( 60,-30 ) ( 30,30 ) ( 30,0 ) ( 0,0)(0,-1)30 ( 0,-30)(1,0)30 ( 30,-30)(1,0)30 ( 60,-30)(-1,2)30 ( 30,30)(-1,-1)30 ( 340,560)(-30,0)(0,0 ) ( 300,80)(-10,-10 ) ( 0,20 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,30 ) ( 30,-15 ) ( 60,0 ) ( 60,30 ) ( 30,45 ) ( -10,27)@xmath20 ( -10,-3)@xmath226 ( 28,-25)@xmath229 ( 65,-3)@xmath220 ( 65,27)@xmath37 ( 28,49)@xmath47 ( 0,30)(0,-1)27 ( 0,30)(2,1)27 ( 0,30)(2,-1)57 ( 60,30)(-2,1)27 ( 60,30)(-2,-1)57 ( 60,30)(0,-1)27 ( 30,45)(0,-1)57 ( 30,45)(0,-1)53 ( 0,0)(2,-1)27 ( 0,0)(2,-1)23 ( 60,0)(-2,-1)27 ( 60,0)(-2,-1)23 ( 30,-15)(2,3)28 ( 30,-15)(2,3)25 ( 30,-15)(2,3)22 ( 30,-15)(-2,3)28 ( 30,-15)(-2,3)25 ( 30,-15)(-2,3)22 ( 70,10 ) ( 60,80)(-10,-10 ) ( 0,20)@xmath230 \left[\begin{array}{rrrrrr } 0&0&1&1&1&-3\\ -1&-1&0&0&0&2 \end{array}\right]\end{array}$ ] ( 30,-20)model iv of @xmath231 ( 270,70 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,-30 ) ( 0,-60 ) ( 30,-60 ) ( 60,-60 ) ( 90,-60 ) ( 30,-30 ) ( 0,0)(0,-1)60 ( 0,-60)(1,0)90 ( 90,-60)(-3,2)90 ( 0,120 ) ( 300,80)(-10,-10 ) ( 0,20 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,30 ) ( 30,-15 ) ( 60,0 ) ( 60,30 ) ( 30,45 ) ( -10,27)@xmath226 ( -10,-3)@xmath47 ( 28,-25)@xmath229 ( 65,-3)@xmath220 ( 65,27)@xmath37 ( 28,49)@xmath20 ( 0,30)(0,-1)27 ( 0,30)(2,-1)57 ( 0,0)(2,1)57 ( 0,0)(2,1)53 ( 30,-15)(2,1)27 ( 30,-15)(-2,1)27 ( 60,0)(0,1)27 ( 60,0)(0,1)23 ( 60,30)(-2,-3)29 ( 60,30)(-1,0)57 ( 60,30)(-2,1)27 ( 60,30)(-2,1)23 ( 30,45)(-2,-1)27 ( 30,45)(0,-1)57 ( 70,10 ) ( 60,80)(-10,-10 ) ( 0,20)@xmath232 \left[\begin{array}{rrrrrr } 0&2&0&0&-1&-1\\ -1&-1&1&1&0&0 \end{array}\right]\end{array}$ ] ( 30,-20)model iii of @xmath231 ( 335,0 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,30 ) ( 0,60 ) ( -30,60 ) ( -60,60 ) ( -60,30 ) ( -30,30 ) ( 0,0)(0,1)60 ( 0,60)(-1,0)60 ( -60,60)(0,-1)30 ( -60,30)(2,-1)60 ( 0,240 ) ( 300,80)(-10,-10 ) ( 0,20 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,30 ) ( 30,-15 ) ( 60,0 ) ( 60,30 ) ( 30,45 ) ( -10,27)@xmath20 ( -10,-3)@xmath47 ( 28,-25)@xmath220 ( 65,-3)@xmath37 ( 65,27)@xmath229 ( 28,49)@xmath226 ( 0,30)(0,-1)27 ( 0,30)(2,-1)57 ( 0,30)(2,-1)53 ( 0,0)(2,3)27 ( 0,0)(2,-1)27 ( 30,-15)(-2,3)27 ( 30,-15)(2,3)27 ( 60,0)(-2,3)27 ( 60,0)(-1,0)57 ( 60,0)(-2,-1)27 ( 60,30)(-1,0)57 ( 60,30)(0,-1)27 ( 30,45)(2,-1)27 ( 30,45)(-2,-1)27 ( 70,10 ) ( 60,80)(-10,-10 ) ( 0,20)@xmath233 \left[\begin{array}{rrrrrr } -1&-1&0&1&1&0\\ 1&-1&-1&0&0&1 \end{array}\right]\end{array}$ ] ( 30,-20)model ii of @xmath231 ( 280,40 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,-30 ) ( 30,-30 ) ( 60,-30 ) ( 60,0 ) ( 30,30 ) ( 30,0 ) ( 0,0)(0,-1)30 ( 0,-30)(1,0)60 ( 60,-30)(0,1)30 ( 60,0)(-1,1)30 ( 30,30)(-1,-1)30 ( 0,360 ) ( 300,80)(-10,-10 ) ( 0,20 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,30 ) ( 30,-15 ) ( 60,0 ) ( 60,30 ) ( 30,45 ) ( -10,27)@xmath20 ( -10,-3)@xmath37 ( 28,-25)@xmath47 ( 65,-3)@xmath220 ( 65,27)@xmath226 ( 28,49)@xmath229 ( 0,30)(0,-1)27 ( 0,0)(2,-1)27 ( 30,-15)(2,1)27 ( 60,0)(0,1)27 ( 60,30)(-2,1)27 ( 30,45)(-2,-1)27 ( 0,0)(1,0)57 ( 60,0)(-2,3)28 ( 30,45)(-2,-3)28 ( 60,30)(-1,0)57 ( 0,30)(2,-3)28 ( 30,-15)(2,3)28 ( 70,10 ) ( 60,80)(-10,-10 ) ( 0,20)@xmath234 \left[\begin{array}{rrrrrr } 0&1&1&0&-1&-1\\ -1&0&1&1&0&-1 \end{array}\right]\end{array}$ ] ( 30,-20)model i of @xmath231 ( 280,30 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,-30 ) ( 30,-30 ) ( 60,0 ) ( 60,30 ) ( 30,30 ) ( 30,0 ) ( 0,0)(0,-1)30 ( 0,-30)(1,0)30 ( 30,-30)(1,1)30 ( 60,0)(0,1)30 ( 60,30)(-1,0)30 ( 30,30)(-1,-1)30 ( 0,480 ) ( 300,80)(-10,-10 ) ( 0,30 ) ( 100,80)(0,0 ) ( 0,0 ) ( 30,0 ) ( 30,-30 ) ( 60,0 ) ( 0,30 ) ( -10,27)@xmath37 ( -10,-3)@xmath20 ( 28,-40)@xmath47 ( 65,-3)@xmath220 ( 28,5)@xmath226 ( 0,0)(0,1)27 ( 0,0)(1,-1)27 ( 30,-30)(1,1)27 ( 30,-30)(0,1)27 ( 30,0)(-1,1)27 ( 30,0)(-1,0)23 ( 30,0)(-1,0)27 ( 0,30)(2,-1)57 ( 0,30)(1,-2)27 ( 60,0)(-1,0)27 ( 60,0)(-1,0)23 ( 70,10 ) ( 60,80)(-10,-10 ) ( 0,20)@xmath235 \left[\begin{array}{rrrrrr } -1&0&1&1&-1\\ -1&-1&0&1&1 \end{array}\right]\end{array}$ ] ( 30,-20)model ii of @xmath228 ( 280,30 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,-30 ) ( 30,-30 ) ( 60,0 ) ( 30,30 ) ( 30,0 ) ( 0,0)(0,-1)30 ( 0,-30)(1,0)30 ( 30,-30)(1,1)30 ( 60,0)(-1,1)30 ( 30,30)(-1,-1)30 ( 340,240)(-30,-10)(0,120 ) ( 120,80)(-10,-10 ) ( 0,0)(1,0)60 ( 60,0)(0,1)60 ( 60,60)(-1,0)30 ( 30,60)(-1,-1)30 ( 0,30)(0,-1)30 ( 0,0 ) ( 30,0 ) ( 60,0 ) ( 60,30 ) ( 60,60 ) ( 30,60 ) ( 0,30 ) ( 30,30 ) ( 140,120 ) ( 120,80)(-10,-10 ) ( 0,0)(1,0)90 ( 90,0)(-1,1)60 ( 30,60)(-1,-1)30 ( 0,30)(0,-1)30 ( 0,0 ) ( 30,0 ) ( 60,0 ) ( 90,0 ) ( 60,30 ) ( 30,60 ) ( 0,30 ) ( 30,30 ) ( 280,120 ) ( 120,80)(-10,-10 ) ( 0,0)(1,0)120 ( 120,0)(-1,1)60 ( 60,60)(-1,-1)60 ( 0,0 ) ( 30,0 ) ( 60,0 ) ( 90,0 ) ( 120,0 ) ( 90,30 ) ( 60,60 ) ( 30,30 ) ( 60,30 ) ( 0,0 ) ( 120,80)(-10,-10 ) ( 0,0)(1,0)60 ( 60,0)(0,1)60 ( 60,60)(-1,0)60 ( 0,60)(0,-1)60 ( 0,0 ) ( 30,0 ) ( 60,0 ) ( 60,30 ) ( 60,60 ) ( 30,60 ) ( 0,60 ) ( 0,30 ) ( 30,30 ) ( 140,0 ) ( 120,80)(-10,-10 ) ( 0,0)(1,0)90 ( 90,0)(-1,1)60 ( 30,60)(-1,0)30 ( 0,60)(0,-1)60 ( 0,0 ) ( 30,0 ) ( 60,0 ) ( 90,0 ) ( 60,30 ) ( 30,60 ) ( 0,60 ) ( 0,30 ) ( 30,30 ) ( 280,0 ) ( 120,80)(-10,-10 ) ( 0,0)(1,0)90 ( 90,0)(-1,1)90 ( 0,90)(0,-1)90 ( 0,0 ) ( 30,0 ) ( 60,0 ) ( 90,0 ) ( 60,30 ) ( 30,60 ) ( 0,90 ) ( 0,60 ) ( 0,30 ) ( 30,30 ) ( 30,105)@xmath236 ( 180,105)@xmath237 ( 330,105)@xmath238 ( 30,-10)@xmath239 ( 180,-10)@xmath240 ( 330,-10)@xmath241 * .. example.*[examples quiver polygon ] figures [ fig : models ] and [ fig : dp3 models ] show some examples of the above relation between quivers , rank @xmath37 subgroups @xmath38 of @xmath39 as in [ introl ] and polygons .
the quivers are taken from @xcite .
we name our examples neutrally as @xmath242 , but also mention the names given to these quiver models in @xcite .
the latter names refer to del pezzo surfaces @xmath243 , i.e. the projective plane @xmath24 blown up in @xmath244 points , some of which , on the present occasion , may happen to be `` infinitely near '' and therefore require repeated blow - ups . with the latter somewhat liberal use of the name _ del pezzo surface _ , @xmath243 matches well with the toric geometry of the fan whose @xmath20-dimensional rays are the half - lines from the interior lattice point @xmath17 through a lattice point @xmath16 at the boundary .
note however that this fan is in general not the same as the secondary fan and that therefore the singularity associated with the polygon ( cf .
@xcite ) is not the singularity obtained by contracting to a point the zero - section of the canonical bundle of the del pezzo surface .
the polygons in figures [ fig : models ] and [ fig : dp3 models ] , however , are in general different from , the polygons associated with the same quiver in @xcite figure 8 .
we will clarify this issue in [ dp3ivtwist ] , [ zigzagquiver ] , [ twist ] .
it is remarkable that the polygons in figures [ fig : models ] and [ fig : dp3 models ] are exactly the lattice polygons with one interior lattice point and @xmath245 lattice points on the boundary .
there are exactly @xmath246 lattice polygons with one interior lattice point .
the remaining @xmath229 are shown in figure [ fig : reflexive rest ] .
a nice review of these so - called _ reflexive polygons _ can be found in @xcite . in cases
@xmath25 , @xmath247 and @xmath248 the greatest common divisor @xmath105 of the entries in the quiver s adjacency matrix is @xmath249 and the polygon in these cases depends on an integer matrix with determinant @xmath105 .
we have chosen this matrix such that the polygon is a reflexive polygon .
* .. * [ introgkz ] in the late 1980 s gelfand , kapranov and zelevinsky discovered fascinating generalizations of the classical hypergeometric structures of euler , gauss , appell , lauricella , horn @xcite .
the main ingredient for these new hypergeometric structures is a finite sequence @xmath250 of vectors in @xmath251 which generates @xmath251 as an abelian group and for which there exists a group homomorphism @xmath252 such that @xmath253 for all @xmath90 .
the latter condition means that @xmath0 lies in a @xmath244-dimensional affine hyperplane in @xmath251 .
figure [ fig : classical examples ] shows @xmath0 ( each black dot represents one vector ) sitting in this hyperplane for some classical hypergeometric structures .
gelfand , kapranov and zelevinsky called these new structures _
@xmath0-hypergeometric systems _ , but nowadays most authors call them _ gkz hypergeometric systems_. let @xmath38 denote the lattice (= free abelian group ) of linear relations in @xmath0 : @xmath254 it follows from the above construction and assumptions that the quotient group @xmath130 is @xmath251 , and a fortiori , it is torsion free . in this paper
we consider only the case when the rank of @xmath38 is @xmath37 ; i.e. @xmath255 . since we assumed the existence of a linear map @xmath256 such that @xmath253 for all @xmath90
, @xmath38 lies in the kernel of the map @xmath257 defined by : @xmath258 the following lemma shows what the other condition for @xmath38 in [ introl ] means in terms of @xmath0 . *
.. proposition.*[lemma : circuits ] * proof .
* assume @xmath38 is contained in the @xmath90-th coordinate hyperplane .
then the @xmath90-th component of every element of @xmath38 is @xmath98 and hence @xmath259 occurs in no linear relation for the vectors in @xmath0 . in other words
, @xmath38 is the lattice of linear relations between the vectors of @xmath260 .
since @xmath38 has rank @xmath37 , we conclude that statement 3 implies statement 1 .
conversely , assume that the lattice @xmath261 of linear relations between the vectors of @xmath260 has rank @xmath249 .
since @xmath261 is contained in @xmath38 and the latter has rank @xmath37 , we see that @xmath262 .
therefore , since @xmath261 is contained in the @xmath90-th coordinate hyperplane , so is @xmath38 .
hence statement 1 implies statement 3 .
it is obvious that statements 1 and 2 are equivalent .
@xmath126 * .. definition.*[def : minrel ] a subsequence @xmath263 of @xmath0 is said to be _ minimally dependent _ if the vectors in @xmath263 are linearly dependent and the vectors in every subsequence @xmath264 of @xmath263 with @xmath265 are linearly independent .
a linear dependence relation @xmath266 is a
_ minimal linear dependence relation _ in @xmath0 if the subsequence @xmath267 is minimally dependent .
* .. proposition.*[prop : circuits ] * proof . * for statement 1 , let @xmath268 be the matrix with columns @xmath269 . then equation ( [ eq : l ] ) can be rewritten as @xmath270 .
this implies @xmath271 , which is just a compressed form of the relations we wanted to prove . for statement 2 note
that condition 3 in proposition [ lemma : circuits ] implies that every linear relation ( [ eq : minrel ] ) is minimal .
conversely , consider a minimal linear relation @xmath266 .
then the subsequence @xmath267 has at most @xmath272 elements .
the given minimal relation is therefore a non - zero scalar multiple of at least one relation ( [ eq : minrel ] ) . @xmath126 * .. * [ gkzsystem ] gelfand , kapranov and zelevinsky @xcite associate with a set @xmath0 as in [ introgkz ] and a vector @xmath273 the following system of partial differential equations for functions @xmath274 of @xmath40 variables @xmath275 : * for every @xmath67 one differential equation @xmath276 * the system of @xmath277 differential equations @xmath278 * .. example.*[cubic roots ] the sequence @xmath279 for example @xmath149 in figure [ fig : models ] can be taken to be @xmath280\,,\quad { \mathbf{a}}_3=\left[\begin{array}{r}1\\ 1\end{array}\right]\,,\quad { \mathbf{a}}_4=\left[\begin{array}{r}1\\ 2\end{array}\right]\,,\quad { \mathbf{a}}_2=\left[\begin{array}{r}1\\ 3\end{array}\right]\,.\ ] ] it is a classical result of k. mayr that the roots of the @xmath20-variable cubic polynomial @xmath281 as functions of the coefficients @xmath282 satisfy the gkz system of differential equations for this @xmath0 and with @xmath283 $ ] ; see e.g. @xcite 2.1 . * .. example.*[labc ] an important example in the toric geometry constructions of sasaki - einstein manifolds is known under the name @xmath284 ; see e.g. @xcite . here
@xmath285 are integers with @xmath286 and @xmath287 the polygon for this example , displayed in @xcite figure 2 , is a quadrangle with vertices @xmath288 , @xmath289 , @xmath290 , @xmath291 where @xmath244 and @xmath292 are integers such that @xmath293 .
the method explained in [ polygon to l ] yields @xmath294 . from this
we see that we can take @xmath279 with @xmath295\,,\quad { \mathbf{a}}_2=\left[\begin{array}{c}1\\ c\end{array}\right]\,,\quad { \mathbf{a}}_3=\left[\begin{array}{c}1\\ 0\end{array}\right]\,,\quad { \mathbf{a}}_4=\left[\begin{array}{c}1\\ b\end{array}\right]\,.\ ] ] as in example [ cubic roots ] this means that the corresponding gkz system deals with the roots of the `` four - nomial '' @xmath296 as functions of the coefficients @xmath297 .
according to @xcite 3 the examples @xmath298 of @xcite are special cases of the above : @xmath299 .
so , these correspond to the `` four - nomials '' + @xmath300 .
thus example [ cubic roots ] is in fact @xmath301 .
( 300,100)(30,10 ) ( 35,20 ) ( 100,80)(0,0 ) ( 0,0)(1,0)40 ( 0,0)(0,1)40 ( 40,40)(-1,0)40 ( 40,40)(0,-1)40 ( 0,0 ) ( 40,0 ) ( 0,40 ) ( 40,40 ) ( 140,20 ) ( 100,80)(0,0 ) ( 0,0)(1,0)40 ( 0,0)(0,1)40 ( 40,40)(-1,0)40 ( 40,40)(0,-1)40 ( 60,20)(0,1)40 ( 40,0)(1,1)20 ( 40,40)(1,1)20 ( 0,40)(3,1)60 ( 0,0)(15,5)4(3,1)10 ( 0,0 ) ( 40,0 ) ( 0,40 ) ( 40,40 ) ( 60,20 ) ( 60,60 ) ( 260,20 ) ( 100,80)(0,0 ) ( 0,30)(1,0)45 ( 45,30)(1,1)22 ( 0,30)(15,15)2(1,1)10 ( 25,55)(15,0)3(1,0)10 ( 0,30)(1,2)35 ( 0,30)(3,-4)35 ( 70,55)(-1,-2)35 ( 69,55)(-3,4)35 ( 46,30)(-1,6)12 ( 46,30)(-1,-4)11 ( 25,55)(6,-36)2(1,-6)3 ( 25,55)(5,20)2(1,4)4 ( 0,30 ) ( 46,30 ) ( 69,55 ) ( 25,55 ) ( 35,-16 ) ( 35,100 ) * .. example.*[f14 ] from the pictures of @xmath0 shown in figure [ fig : classical examples ] one easily sees that for the gauss system the lattice @xmath38 is generated by the vector @xmath302 and thus has rank @xmath20 and is not of interest here . for appell s @xmath303 and @xmath304 the lattice @xmath38 has rank @xmath37 and the corresponding @xmath82-matrices are @xmath305 \,,\quad f_4:\ ; \left[\begin{array}{rrrrrr}-1&-1&1&1&0&0\\-1&-1&0&0&1&1\end{array}\right].\ ] ] thus appell s @xmath303 corresponds to @xmath306 in figure [ fig : dp3 models ] . for appell
s @xmath304 the quiver and the polygon are ( 300,90)(-10,-10 ) ( 80,20 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,30 ) ( 30,-15 ) ( 60,0 ) ( 60,30 ) ( 30,45 ) ( -10,27)@xmath20 ( -11,-3)@xmath229 ( 28,-28)@xmath220 ( 65,-3)@xmath37 ( 65,27)@xmath226 ( 28,49)@xmath47 ( 0,30)(0,-1)27 ( 0,0)(2,-1)27 ( 30,-15)(2,1)27 ( 60,0)(0,1)27 ( 60,30)(-2,1)27 ( 30,45)(-2,-1)27 ( 0,30)(1,0)57 ( 0,0)(2,3)28 ( 30,45)(2,-3)28 ( 60,0)(-1,0)57 ( 60,30)(-2,-3)28 ( 30,-15)(-2,3)28 ( 250,30 ) ( 100,80)(0,0 ) ( 0,-30 ) ( 30,-30 ) ( 60,-30 ) ( 60,0 ) ( 60,30 ) ( 30,0 ) ( 0,-30)(1,0)60 ( 60,-30)(0,1)60 ( 60,30)(-1,-1)60 in @xcite 3.2 gelfand , kapranov and zelevinsky discuss how the @xmath307 complete hypergeometric horn series in two variables fit their theory .
it is an amusing simple exercise to now find the corresponding quivers and polygons .
it turns out that , with the exception of @xmath304 , all horn series give a quiver listed in example [ examples quiver polygon ] : @xmath308 * .. * [ l to a ] in [ introgkz ] we started from the sequence of vectors @xmath0 and then defined @xmath38 via equation ( [ eq : l ] ) .
let us reverse the procedure and start with a rank @xmath37 subgroup @xmath41 as in [ introl ] .
we can now define @xmath309 if the quotient group @xmath130 is torsion free , it is isomorphic to @xmath310 and the sequence of vectors @xmath0 in ( [ eq : afroml ] ) satisfies the requirements of [ introgkz ] .
_ thus ( [ eq : l ] ) and ( [ eq : afroml ] ) give equivalences , inverse to each other , between the data @xmath0 and @xmath38 . _ in combination with theorem [ quivergrassmann ]
this gives * .. theorem.*[quiverhypergeo ] _ there is an equivalence of data between on the one hand sequences @xmath0 as in [ introgkz ] which satisfy also the conditions in proposition [ lemma : circuits ] and on the other hand quivers without isolated nodes or directed loops of length @xmath92 , such that in every node the number of incoming arrows equals the number of outgoing arrows and such that for its anti - symmetrized adjacency matrix the rank is @xmath37 and the greatest common divisor of the entries is @xmath20 .
_ @xmath126 * .. remark.*[gkzwithtorsion ] we refer to [ gkz with torsion ] for a discussion of how to extend the definition of gkz systems so as to allow for torsion in the group @xmath130 . * .. * [ a secondary ] in order to have an efficient method to construct bases for the solution space of the hypergeometric system ( [ eq : gkz1])-([eq : gkz2 ] ) gelfand , kapranov and zelevinsky developed the theory of the _ secondary fan _ and the _ secondary polytope_. the term `` secondary '' refers to the habit of considering the set @xmath0 as primary data and to call the convex hull of @xmath0 the _ primary polytope_. in case the lattice @xmath38 of relations in @xmath0 has rank @xmath37 , the constructions become particularly simple .
the secondary fan is the one in definition [ def : lsecondary fan ] . with a @xmath37-dimensional cone @xmath101 in the secondary fan one
associates a set @xmath146 of @xmath37-element subsets of @xmath142 as in equation ( [ eq : lc ] ) .
one can interpret this @xmath146 is as a triangulation of the primary polytope @xmath311 as follows .
recall that @xmath250 and assume for convenience of imagining pictures that all @xmath259 s are different .
for @xmath216 set @xmath312 .
this @xmath313 is an @xmath314-dimensional simplex and the simplices @xmath313 with @xmath216 together constitute a _ triangulation the primary polytope_. in @xcite p.220 the _ secondary polytope _
is constructed ( and defined ) in terms of these triangulations as follows . for a @xmath37-element subset @xmath315 with @xmath316
let @xmath317 denote the @xmath318-matrix with columns @xmath319 for @xmath320 in the natural order of increasing indices .
to a @xmath37-dimensional cone @xmath101 in the secondary fan one then associates the vector @xmath321 then in @xcite p.220 the _ secondary polytope _ @xmath322 is defined as @xmath323 as @xmath324 is @xmath325 times the euclidean volume of the simplex @xmath313 , the number @xmath326 is for every @xmath101 equal to @xmath325 times the euclidean volume of the primary polytope @xmath311 and the point @xmath327 coincides for every @xmath101 with the barycentre of the primary polytope .
the secondary polytope therefore lies in a @xmath37-dimensional plane parallel to @xmath328 . * .. * [ compare secondary polytopes ] let us compare @xmath322 with the polytope @xmath147 of definition [ def : lsecondary polytope ] .
it is a well - known and easy to prove fact that @xmath329 for all @xmath216 ( see e.g. @xcite eq .
thus equations ( [ eq : phic ] ) and ( [ eq : psic ] ) yield @xmath330 with @xmath331 this means that the two secondary polytopes @xmath332 and @xmath144 are related by a point symmetry with centre @xmath333 : @xmath334
* .. * [ gkztorusaction ] the variables in gkz theory ( see section [ sec : hyper ] ) are the natural coordinates on the space @xmath335 of maps from @xmath0 to @xmath336 .
the torus @xmath337 of group homomorphisms from @xmath251 to @xmath338 , acts naturally on @xmath339 and on the functions on this space : for @xmath340 , @xmath341 , @xmath342 and @xmath343 : @xmath344 the gkz hypergeometric functions associated with @xmath0 and @xmath345 are defined on open domains in @xmath339 .
one easily sees that if a function @xmath274 on @xmath339 satisfies the differential equations ( [ eq : gkz1 ] ) then for every @xmath340 the function @xmath346 also satisfies these differential equations .
so , the torus @xmath347 acts on the solution space of the system of differential equations ( [ eq : gkz1 ] ) . on the other hand ,
if @xmath348 and @xmath349 are two functions which satisfy the differential equations ( [ eq : gkz2 ] ) with the same @xmath345 , then their quotient @xmath350 satisfies @xmath351 the latter equation is equivalent to @xmath352 being @xmath347-invariant .
thus we find : _ all quotients of pairs of solutions of the system ( [ eq : gkz1])-([eq : gkz2 ] ) are functions on simply connected open subsets of the orbit space @xmath353 . in the case of interest in the present paper
the dimension of this orbit space is @xmath354 . _ for @xmath355 the differential equations ( [ eq : gkz2 ] )
are equivalent with @xmath274 transforming under the action of @xmath347 according to the character given by @xmath345 : @xmath356 in particular for @xmath357 all solutions of ( [ eq : gkz2 ] ) are @xmath347-invariant . * .. * [ toric sec fan ] the space @xmath358 of maps from @xmath0 to @xmath338 is a torus of dimension @xmath40 which contains @xmath347 as a subtorus .
the action ( [ eq : torus action on functions ] ) of @xmath340 on @xmath339 obviously restricts to the action of @xmath347 on @xmath359 by multiplication .
the quotient space @xmath360 , which is a subspace of @xmath353 , is the torus @xmath361 of group homomorphisms from @xmath38 to @xmath338 .
the toric variety associated with the secondary fan ( cf .
definition [ def : lsecondary fan ] ) gives a compactification of the torus @xmath361 ; see @xcite for the general theory of toric varieties .
it is an essential part of the gkz philosophy that quotients of hypergeometric functions should be viewed as being defined on open subsets of this toric variety . * .. theorem.*[thm : solution dimension ] ( cf .
@xcite theorems 2 and 5 , @xmath362 , @xcite prop .
13.5 ) _ let _
@xmath363 _ assume @xmath364 , then the dimension of the space of solutions of the system of differential equations ( [ eq : gkz1])-([eq : gkz2 ] ) at a general point of @xmath339 equals the number @xmath365 in equation ( [ eq : vola ] ) . _
@xmath126 * .. remark.*[unimodular ] as shown in the proof of @xcite prop . 13.15 the condition @xmath364 in the above theorem is satisfied if @xmath0 admits a _
unimodular triangulation_. the latter condition is equivalent to : there is a cone @xmath101 in the secondary fan such that @xmath366 for all @xmath216 ( see definitions [ def : lsecondary fan ] and [ def : lsecondary polytope ] ) . * .. remark.*[gkz with torsion ] from the discussion in [ gkztorusaction ] one may get an idea about extending the gkz system ( [ eq : gkz1])-([eq : gkz2 ] ) to the situation in which @xmath130 has a non - trivial torsion subgroup @xmath367 .
equation ( [ eq : gkz1 ] ) still makes sense in this more general situation , but equation ( [ eq : gkz2 ] ) must be adapted .
recall from ( [ eq : afroml ] ) a definition of @xmath0 which also works in the torsion case .
let @xmath368 .
this is a commutative algebraic group of which the connected component of the identity is @xmath369 and the group of connected components is the finite abelian group @xmath370 .
formula ( [ eq : torus action on functions ] ) defines an action of @xmath371 on @xmath339 and on the functions on @xmath339 .
it is now clear how to adapt equation ( [ eq : gkz2 ] ) : let @xmath372 denote the projection of @xmath259 in the free part @xmath251 of @xmath130 and replace ( [ eq : gkz2 ] ) by @xmath373 plus the requirement that the solution should transform according to some character of the finite abelian group @xmath367 .
* .. example.*[example b1 ] in case @xmath25 in figure [ fig : models ] one has @xmath374 and @xmath375 . as a generator for the torsion subgroup we take @xmath376 . then the polynomials @xmath377 , @xmath378 and @xmath379 satisfy the differential equations ( [ eq : gkz1 ] ) for @xmath38 , the differential equations ( [ eq : gkz2 ] ) for @xmath380 , while @xmath381 for @xmath382 .
we are going to describe a construction which associates with a rank @xmath37 subgroup @xmath38 of @xmath39 , as in [ introl ] , _ dessins denfants _ , i.e. bipartite graphs embedded in oriented riemann surfaces .
the construction is given in @xcite 5 and 6 as _ fast inverse algorithm_. in op .
, however , this algorithm is only presented via explicit visual inspection of pictures in some concrete examples . in this section
we want to present a general principle behind the fast inverse algorithm of @xcite which uses only linear algebra and can be performed by computer .
the main part of our construction produces rhombus tilings of the plane and is the same as n.g .
de bruijn s @xcite construction of penrose tilings .
his work also led to effective methods for making _ quasi - crystals _ @xcite . since @xmath38 is defined over @xmath74 the construction yields in our situation periodic tilings of the plane @xmath383 , not just quasi - periodic ones .
we can therefore pass to @xmath191 modulo the period lattice and find a tiling of the two - dimensional torus .
* .. * [ grids ] in this section @xmath38 is as in [ introl ] .
the quotient group @xmath130 is allowed to have torsion .
let @xmath328 denote the real @xmath37-plane in @xmath45 which contains @xmath38 .
we write @xmath384 for the real @xmath37-plane in @xmath45 obtained by translating @xmath328 over a vector @xmath385 . on the other hand one
has in @xmath45 the _ standard @xmath40-grid _ consisting of the hyperplanes @xmath386 by intersecting with this standard @xmath40-grid we obtain in the @xmath37-plane @xmath384 an @xmath40-grid of lines : @xmath387 note that this crucially uses the assumption that @xmath38 is not contained in any of the standard coordinate hyperplanes in @xmath39 . _ the grid lines are naturally oriented _ :
the orientation on @xmath388 is such that the points in @xmath384 which are to the left of @xmath388 , have a larger @xmath90-th coordinate than the points to the right of @xmath388 . * .. definition.*[def : nonresonant ] we say that @xmath389 is _ non - resonant _ if no three grid lines pass through one point . * .. first step.*[first step ] let @xmath38 and @xmath389 be as in [ grids ] with @xmath389 non - resonant as in [ def : nonresonant ] .
consider the map @xmath390 where @xmath391 for a real number @xmath392 denotes the largest integer @xmath393 .
the map @xmath110 contracts an open @xmath40-cube in the @xmath39 structure on @xmath45 ( and part of its boundary ) onto one of its corners .
@xmath110 commutes with the translation action of @xmath39 on @xmath45 and @xmath39 .
suppressing @xmath38 and @xmath389 from the notation we define @xmath394 the map @xmath110 is constant on each @xmath37-cell ( = connected component ) of the grid complement in @xmath384 .
every point of @xmath395 is in fact the image of a unique such @xmath37-cell .
the distance between two points of @xmath395 is @xmath20 if and only if the corresponding @xmath37-cells are separated by exactly one grid line . because @xmath389 is non - resonant , an intersection point of grid lines is in the closure of exactly four @xmath37-cells .
if @xmath396 the four points of @xmath395 corresponding to these cells , i.e. @xmath397 , @xmath398 , @xmath399 , @xmath400 , are the vertices of a square @xmath401 .
when @xmath402 runs through the set of all intersection points of grid lines , these squares fit together to a connected surface @xmath403 embedded in @xmath45 .
one can characterize the surface @xmath404 also as @xmath405 here one may define _ unit square _ as the convex hull of four points @xmath406 in @xmath45 with euclidean distances @xmath407 @xmath408 and @xmath409 .
a unit square with vertices in @xmath39 is necessarily of the form @xmath410 for some @xmath411 and some @xmath412 .
we say that a unit square in @xmath45 has _ type @xmath96 _ if its sides are parallel to the vectors @xmath133 or @xmath413 .
similarly , a side of a unit square in @xmath45 has _ type @xmath90 _ if it is parallel to the vector @xmath133 . * .. proposition.*[prop : sprojection ] _ let @xmath414 denote the real dual of @xmath38 .
identify the real dual @xmath415 of @xmath39 with @xmath45 by means of the standard dot product . then
the linear map @xmath416 which is dual to the inclusion @xmath417 , restricts to a homeomorphism @xmath418
. _ * proof .
* let @xmath419 denote the images of the standard basis vectors of @xmath45 .
after identifying in the obvious way , the real plane @xmath328 with the real @xmath37-plane @xmath384 in @xmath45 one obtains in @xmath328 an @xmath40-grid in which the grid lines @xmath388 are perpendicular to @xmath190 . after choosing a basis for @xmath38 and taking the dual basis for @xmath136 we may identify @xmath328 and @xmath136 with @xmath191 .
the term `` perpendicular '' then means perpendicular with respect to the standard inner product on @xmath191 . next draw for each intersection point of grid lines , say @xmath396 , a parallelogram with centre @xmath402 and sides @xmath420 and @xmath421 . here
the positive real number @xmath422 is so small that the parallelograms obtained from all grid intersection points are disjoint ; see figure [ fig : grid2tiling ] for an example .
it is clear from this figure how one can glue the parallelograms of two consecutive intersection points on a grid line along their sides perpendicular to the grid line .
the result of this glueing is then the same as the image ( scaled with a factor @xmath422 ) of the surface @xmath404 under the projection @xmath134 . @xmath126
( 350,130)(0,0 ) ( -20,0 ) ( 100,100)(0,15 ) ( 180,0 ) ( 150,100)(0,0 ) it is obvious from the construction that the group @xmath38 acts by translations on the point set @xmath395 and on the surface @xmath404 , preserving the types of the squares and their sides . the cell structure on @xmath404 given by the squares , their sides and vertices induces therefore on @xmath423 a cell structure .
the following proposition counts the cells on @xmath423 . *
.. proposition.*[count cells ] * proof .
* the counts can be made on the dual structure , which is given by the intersecting grid lines on @xmath328 .
let @xmath424 denote the @xmath115-matrix with columns @xmath190 and @xmath425 . then 1 .
amounts to counting for @xmath426 the number of elements in the set @xmath427 .
since there is a bijection between this set and the coset space @xmath428 , the number of elements is @xmath429 .
the number of @xmath20-cells of type @xmath90 is equal to the number of intersection points of the @xmath90-th grid with the other grids . in view of 1 .
this number is therefore as stated in 2 . knowing the numbers of @xmath98- and @xmath20-cells on the @xmath37-torus @xmath430 one computes the number of @xmath37-cells from the fact that the euler characteristic of a @xmath37-torus is @xmath98 .
@xmath126 * .. second step.*[deforms ] for the constructions in [ grids ] and [ first step ] we may randomly choose a non - resonant @xmath385 .
once we have constructed the set @xmath395 and surface @xmath404 we modify it by _ elementary transformations_. for such an elementary transformation we need a point @xmath431 such that there are exactly three points @xmath432 in @xmath395 at distance @xmath20 from @xmath433 and such that the three points @xmath434 , @xmath435 and @xmath436 also lie in @xmath395 .
these seven points are the vertices of three squares in @xmath404 .
the elementary transformation now replaces @xmath433 by @xmath437 and replaces the three squares by the three squares in @xmath45 with vertex configuration @xmath438 @xmath439 see figure [ fig : elementary transformation ] .
( 300,90)(-20,-10 ) ( 35,20 ) ( 100,80)(0,0 ) ( 0,0)(1,1)30 ( 0,0)(1,-1)30 ( 60,0)(-1,1)30 ( 60,0)(-1,-1)30 ( 60,0)(1,0)42 ( 30,30)(1,0)42 ( 30,-30)(1,0)42 ( 102,0)(-1,1)30 ( 102,0)(-1,-1)30 ( 0,0 ) ( 60,0 ) ( 102,0 ) ( 30,30 ) ( 30,-30 ) ( 72,30 ) ( 72,-30 ) ( 180,20)@xmath440 ( 210,20 ) ( 100,80)(0,0 ) ( 30,30)(-1,-1)30 ( 30,-30)(-1,1)30 ( 60,0)(-1,-1)30 ( 60,0)(-1,1)30 ( 60,0)(1,0)42 ( 30,30)(1,0)42 ( 30,-30)(1,0)42 ( 102,0)(-1,1)30 ( 102,0)(-1,-1)30 ( 72,30)(-1,-1)30 ( 72,-30)(-1,1)30 ( 0,0)(1,0)42 ( 360,20)@xmath440 ( 390,20 ) ( 100,80)(0,0 ) ( 0,0)(1,1)30 ( 0,0)(1,-1)30 ( 42,0)(1,1)30 ( 42,0)(1,-1)30 ( 42,0)(-1,0)42 ( 30,30)(1,0)42 ( 30,-30)(1,0)42 ( 102,0)(-1,1)30 ( 102,0)(-1,-1)30 ( 0,0 ) ( 42,0 ) ( 102,0 ) ( 30,30 ) ( 30,-30 ) ( 72,30 ) ( 72,-30 ) in order to preserve @xmath38-periodicity we perform this transformation simultaneously at all configurations @xmath441 @xmath442 with @xmath443 .
after this elementary transformation we obtain a surface @xmath444 in @xmath45 which is the union of all unit squares with vertices in the set @xmath445 , obtained from @xmath395 by replacing the points @xmath446 by @xmath447 for all @xmath443 . the surface @xmath444 , in turn , can be further modified by elementary transformations : just replace in the above construction @xmath395 by @xmath445 and @xmath433 by an appropriate point of @xmath445 . *
.. third step.*[specials ] after starting the construction in [ first step ] with a randomly chosen non - resonant @xmath385 one can create by repeated elementary transformations many surfaces @xmath404 each of which is the union of unit squares with vertices in an @xmath38-invariant subset @xmath395 of @xmath39 and which is mapped homeomorphically onto @xmath136 by the projection @xmath134 .
moreover the numbers of cells on each of these surfaces are still the same as in proposition [ count cells ] .
the number of surfaces one can make in this way is finite and it is possible to generate ( by computer ) a complete list . * .. remark.*[start independent ] the list of surfaces produced in [ specials ] is , up to permutation of its entries , independent of the choice of @xmath385 at the start of the algorithm .
this can be seen as follows .
fix @xmath90 , @xmath448 .
take @xmath385 and let @xmath449 for @xmath450 .
set @xmath451 and @xmath452 .
then the surface @xmath453 constructed from @xmath454 in [ first step ] is equal to the surface @xmath455 for @xmath456 .
figure [ fig : grid change ] shows how the grid locally changes as @xmath457 passes through @xmath458 .
( 300,90)(-50,10 ) ( 35,20 ) ( 100,80)(0,0 ) ( 0,0)(1,1)80 ( 80,0)(-1,1)80 ( 50,0)(0,1)80 ( 180,40)@xmath440 ( 210,20 ) ( 100,80)(0,0 ) ( 0,0)(1,1)80 ( 80,0)(-1,1)80 ( 40,0)(0,1)80 ( 360,40)@xmath440 ( 390,20 ) ( 100,80)(0,0 ) ( 0,0)(1,1)80 ( 80,0)(-1,1)80 ( 30,0)(0,1)80 comparing figures [ fig : grid change ] and [ fig : elementary transformation ] one sees that such a local change of the grid corresponds to an elementary transformation of the surface @xmath455 . thus the surfaces @xmath453 with @xmath459 are obtained from @xmath455 by a number of elementary transformations .
since any two non - resonant @xmath460 can be moved to a common value by coordinatewise changes as above we conclude that the surfaces @xmath404 and @xmath444 obtained in [ first step ] from @xmath389 and @xmath461 , respectively , are related by elementary transformations , and that , hence , the lists of surfaces which the algorithm produces from start values @xmath389 and @xmath461 are the same , up to possibly a permutation of the entries . * .. definition.*[perfect surface ]
we say that a surface @xmath404 on this list is _ perfect _ if the function @xmath462 , @xmath463 , takes only three values on the set @xmath395 of vertices in @xmath404 . in that case
these three values are consecutive integers , say @xmath464 .
we denote by @xmath465 ( resp .
@xmath466 resp .
@xmath467 ) the set of vertices where @xmath257 takes the value @xmath468 ( resp
. @xmath469 resp .
@xmath470 ) and say that the vertices in @xmath465 are _ black _ , those in @xmath467 are _ white _ and those in @xmath466 are _
grey_. each of these sets is invariant under the translation action of @xmath38 . * .. remark.*[more perfect surfaces ] i have at present no proof that for every rank @xmath37 subgroup @xmath41 as in [ introl ] the above construction indeed yields at least one perfect surface .
on the other hand , in all examples i investigated the computer produced at least one perfect surface .
it would also be very interesting to know which perfect surfaces can arise already in the first step [ first step ] of the construction , by choosing @xmath389 appropriately ( instead of at random ) .
* .. * [ diagonals ] for every surface @xmath404 in the list of [ specials ] one has the graph @xmath471 with set of vertices @xmath395 and arrows given by the diagonals in the squares , oriented such that in a square of type @xmath96 the oriented diagonals are @xmath472 and @xmath473 .
the advantage of having @xmath404 embedded in @xmath45 is that the vertices and the arrows in these graphs are actual points and vectors in @xmath45 . if the surface @xmath404 is perfect , every square in @xmath404 has one vertex in @xmath465 , one in @xmath467 and two vertices in @xmath466 .
one of its diagonals goes from the white to the black vertex and the other diagonal connects the two grey vertices .
the graph @xmath471 therefore is the disjoint union of two oriented graphs @xmath474 and @xmath475 , with vertex sets @xmath476 and @xmath466 , respectively .
the graph @xmath474 is a _ bi - partite graph _ , i.e. its vertex set is the disjoint union of two sets ( the black resp .
white vertices ) and with edges connecting only vertices of different colors .
there is a natural duality between @xmath474 and @xmath475 : every vertex of @xmath474 lies in a unique connected component of @xmath477 and vice versa .
moreover the boundaries of these connected components are polygons and an arrow between two nodes in one graph is a common side of the corresponding polygons for the other graph . * .. * [ periodic models 1 ] the perfect surface @xmath404 is mapped homeomorphically onto @xmath136 by the projection @xmath134 , which is dual to the inclusion @xmath135 .
this projection maps @xmath133 to @xmath190 , for @xmath139 .
one may however pre - compose this projection with the linear map @xmath478 .
the composite linear map @xmath134 maps the tiling by squares on the surface @xmath404 piecewise linearly and homeomorphically onto a tiling of @xmath136 by rhombi .
it maps @xmath38 isomorphically onto a lattice @xmath479 in @xmath136 and it maps the graphs @xmath474 and @xmath475 ` isomorphically ' onto @xmath479-periodic graphs @xmath480 and @xmath481 in the plane @xmath136 . in the literature @xmath482 is called a _ periodic dimer model _ or _ periodic brane tiling _ ; see for instance @xcite .
* .. definition.*[graphs ] all structures in [ diagonals ] are invariant under translation by vectors in @xmath38 .
passing to the orbit space we obtain from a perfect surface @xmath404 the @xmath37-dimensional torus @xmath483 ( i.e. compact oriented surface of genus @xmath20 without boundary ) .
embedded in this torus are the two graphs @xmath484 and @xmath485 , which are dual to each other .
note that [ periodic models 1 ] also yields @xmath486 , @xmath487 and @xmath488 .
after grothendieck one calls @xmath489 a _ dessin denfants _ @xcite . * .. example.*[dp3ivtwist ] figure [ fig : dp3ivtwist ] shows two drawings of the dessin @xmath489 for @xmath490 from figure [ fig : models ] ; or , rather , it shows a lifting of this dessin to the plane @xmath136 with basis so that the period lattice @xmath479 becomes just @xmath192 .
it also shows @xmath491 as an unembedded quiver .
_ note however that this is not the same as the quiver @xmath1 in case @xmath490 in figure [ fig : models ] .
_ the quiver in figure [ fig : dp3ivtwist ] is in fact the quiver for the singularity @xmath492 where a generator of the cyclic group @xmath493 acts on @xmath494 as multiplication by the diagonal matrix @xmath495 .
also the brane tiling picture in figure [ fig : dp3ivtwist ] appears in @xcite 3 in connection with the singularity @xmath492 .
this difference between @xcite and our approach comes , because we still have to perform the untwist , as is explained below .
( 350,110)(-40,10 ) ( -40,10 ) ( 150,150)(0,0 ) ( 100,10 ) ( 150,150)(0,0 ) ( 250,50 ) ( 100,80)(0,0 ) ( 0,0 ) ( 0,30 ) ( 30,-15 ) ( 60,0 ) ( 60,30 ) ( 30,45 ) ( 0,30)(0,-1)27 ( 0,0)(2,-1)27 ( 30,-15)(2,1)27 ( 60,0)(0,1)27 ( 60,30)(-2,1)27 ( 30,45)(-2,-1)27 ( 0,0)(1,0)57 ( 60,0)(-2,3)28 ( 30,45)(-2,-3)28 ( 60,30)(-1,0)57 ( 0,30)(2,-3)28 ( 30,-15)(2,3)28 ( 30,43)(0,-1)55 ( 30,-13)(0,1)55 ( 2,29)(2,-1)55 ( 58,1)(-2,1)55 ( 2,1)(2,1)55 ( 58,29)(-2,-1)55 * .. * [ zigzagquiver ] the quiver @xmath491 in [ graphs ] is not the plcker quiver of @xmath38 , defined in [ pluckerquiver ]
. the latter appears in the current setting through the _ zigzag loops _
, as follows .
let @xmath404 be a perfect surface .
fix @xmath496 . in every square on @xmath404 which has two sides parallel to @xmath133 draw the line segment connecting the midpoints of these two sides .
if the square has type @xmath96 this line segment is oriented so that it is a translate of the vector @xmath497 .
the union of these line segments projects to an oriented closed curve on @xmath498 , called the _ @xmath90-th zigzag loop_. note that this zigzag loop may consist of several connected components .
according to proposition [ count cells ] the @xmath90-th and @xmath91-th zigzag loops on @xmath498 intersect in exactly @xmath499 points .
_ thus we see that the nodes of the plcker quiver can be identified with the zigzag loops on @xmath498 and the arrows between two nodes are the intersection points of the corresponding zigzag loops .
_ the orientation of the arrows follows from the orientation of the zigzag loops and can in pictures be indicated as an over / undercrossing of the zigzag loops .
figure [ fig : dp3ivzigzag ] shows the zigzag loops for the dessin in figure [ fig : dp3ivtwist ] .
( 350,100)(-40,20 ) ( 40,10 ) ( 150,150)(0,0 ) * .. final untwist.*[twist ] pictures like figure [ fig : dp3ivzigzag ] can be viewed as showing an oriented surface with boundary , in which the connected components of the zigzag loops are the connected components of the boundary and which is embedded in three space in a twisted way so that the black dots are on one side of the surface and the white dots are on the other side .
this surface comes with a tiling by helices as shown in figure [ fig : untwist helix ] .
( 350,200)(-100,-20 ) ( 0,0 ) ( 150,100)(0,0 ) ( 0,0)(1,1)80 ( 0,0)(-1,1)80 ( 0,160)(-1,-1)80 ( 0,160)(1,-1)80 ( 0,0 ) ( 0,160 ) ( 0,0)(1,1)30 ( 0,0)(-1,1)30 ( -80,80)(1,1)30 ( 80,80)(-1,1)30 ( -10,20)(1,0)20 ( -10,140)(1,0)20 ( -12,175)@xmath500 ( -12,-25)@xmath501 ( 45,120)@xmath502 ( -60,120)@xmath503 ( 45,34)@xmath503 ( -60,34)@xmath502 ( 200,0 ) ( 150,100)(0,0 ) ( 0,0)(1,1)40 ( 0,0)(-1,1)40 ( 0,160)(-1,-1)40 ( 0,160)(1,-1)40 ( -40,40)(1,1)80 ( 40,40)(-1,1)35 ( -40,120)(1,-1)35 ( -40,40)(1,1)35 ( 0,0)(1,1)30 ( 0,0)(-1,1)30 ( -40,120)(1,1)30 ( 40,120)(-1,1)30 ( 0,0 ) ( 0,160 ) ( -10,20)(1,0)20 ( -10,140)(1,0)20 ( -12,175)@xmath500 ( -12,-25)@xmath501 ( 45,120)@xmath502 ( -60,120)@xmath503 ( 45,34)@xmath503 ( -60,34)@xmath502 ( 360,0 ) ( 150,100)(0,0 ) ( 0,0)(1,1)40 ( 0,0)(-1,1)40 ( 0,160)(-1,-1)40 ( 0,160)(1,-1)40 ( 40,40)(0,1)80 ( -40,40)(0,1)80 ( 40,40)(0,1)40 ( -40,120)(0,-1)60 ( 0,0)(1,1)30 ( 0,0)(-1,1)30 ( -40,120)(1,1)30 ( 40,120)(-1,1)30 ( 0,0 ) ( 0,160 ) ( 10,20)(-1,0)20 ( -10,140)(1,0)20 ( -12,175)@xmath500 ( -12,-25)@xmath501 ( 45,80)@xmath502 ( -60,80)@xmath503 ( 560,0 ) ( 150,100)(0,0 ) ( 0,0)(1,1)80 ( 0,0)(-1,1)80 ( 0,160)(-1,-1)80 ( 0,160)(1,-1)80 ( 0,0 ) ( 0,160 ) ( 10,20)(-1,0)20 ( -10,140)(1,0)20 ( -60,90)(0,-1)20 ( 60,70)(0,1)20 ( -40,40)(1,1)80 ( 40,40)(-1,1)80 ( 0,80)(1,1)20 ( 0,80)(-1,-1)20 ( 40,40)(-1,1)20 ( -40,120)(1,-1)20 ( -12,175)@xmath500 ( -12,-25)@xmath501 ( 85,80)@xmath502 ( -100,80)@xmath503 one can ` untwist ' the helix tiles and complete the untwisted helix tiles to squares as indicated in figure [ fig : untwist helix ] .
the effect is that the boundary cycles are being capped off by discs .
the result is an oriented surface @xmath5 without boundary which is tiled by squares with one black , one white and two red vertices .
the black - white diagonals give an embeding of the bipartite graph @xmath18 into @xmath5 .
the red - red diagonals give an embeding of the quiver @xmath1 into @xmath5 ( if all zigzag loops have just one connected component ; otherwise some points on the surface must be pinched together ) .
the untwisting procedure is described in @xcite 5 via examples , pictures and visual inspection . here
we want to build the ` untwisted surface ' by simply reinterpreting the combinatorial data of the tiling of the torus @xmath498 , i.e. the @xmath38-periodic tiling of the perfect surface @xmath404 by unit squares taken modulo @xmath38 ( see [ perfect surface ] , [ graphs ] ) .
let @xmath504 ( resp .
@xmath505 ) denote the set of black ( resp .
white ) vertices .
there is a bijection between the set of tiles and the set @xmath2 of arrows of the quiver @xmath1 and we can refer to tiles as @xmath6 .
a tile @xmath23 has one black vertex @xmath30 and one white vertex @xmath31 .
a tile @xmath23 is the image of a unit square with sides parallel to two of the basis vectors @xmath69 ; let us denote the indices of this ( unordered ) pair of basis vectors as a @xmath37-element subset @xmath506 of @xmath142 .
thus the tiling on @xmath498 yields the list of quadruples @xmath507 now , for every @xmath6 we take a unit square @xmath508 and attach labels @xmath30 ,
@xmath31 , @xmath509 , @xmath510 and colors black , white , red , red , respectively , to its vertices such that the vertices labeled @xmath30 and @xmath31 are not adjacent .
if two such squares @xmath508 and @xmath511 have equal labels on two adjacent vertices , we glue @xmath508 and @xmath511 along the corresponding sides .
the result of this glueing is a surface @xmath512 tiled with squares . on the surface @xmath512 there is for every @xmath513 one black point with label @xmath500 and for every @xmath514 there is one white point with label @xmath501 .
however for @xmath66 there may be several red points with label @xmath90 .
we finally obtain the desired surface @xmath5 by identifying for every @xmath66 the red points with label @xmath90 . * .. proposition.*[oriented m ] _ the surface @xmath512 , and hence also @xmath5 , is oriented .
_ * proof . * cut the square @xmath508 in four pieces like in the right - hand picture of figure [ fig : untwist helix ] .
then for fixed @xmath513 the small squares with black vertex @xmath500 are glued together to a polygon isomorphic to the polygon in @xmath498 formed by the unit squares with one vertex @xmath500 .
similarly , for fixed @xmath514 the small squares with white vertex @xmath501 are glued together to a polygon isomorphic to the polygon in @xmath498 formed by the unit squares with one vertex @xmath501 but with the orientation reversed .
for @xmath66 the small squares with red vertex @xmath90 are glued together to a number of disjoint polygons , one for every connected component of the @xmath90-th zigzag loop and with oriented boundary ` equal to ' that component .
it is now clear from figure [ fig : untwist helix ] that these polygons are oriented consistently so that @xmath512 is oriented .
@xmath126 * .. conclusion.*[final m ] _ the surface @xmath5 is tiled with squares so that the black - white diagonals form a bi - partite graph isomorphic to the bi - partite graph @xmath18 in [ graphs ] , while the red - red diagonals form a graph isomorphic to the quiver @xmath1 .
_ * .. example.*[correct dp3iv ] the list @xmath515 ( [ eq : quadruples ] ) which our algorithm produces for model iv of @xmath231 , i.e. example @xmath490 from figure [ fig : models ] , is @xmath516 the surface @xmath5 with the corresponding tiling by kites ( instead of squares ) is shown in the left - hand picture in figure [ fig : dp3iv ] .
the right - hand picture in figure [ fig : dp3iv ] shows the bi - partite graph @xmath18 and the quiver @xmath1 embedded in @xmath5 .
note that this quiver is indeed the same as the one for example @xmath490 in figure [ fig : dp3iv ] .
( 350,150)(-40,-10 ) ( 0,0 ) ( 150,100)(0,15 ) ( 180,0 ) ( 150,100)(0,15 )
* .. definition.*[def : perfect matching ] ( see e.g. @xcite ) a _ perfect matching
_ @xmath517 on a bi - partite graph @xmath18 ( also known as a _ dimer configuration _ ) is a subset of the edges of @xmath18 such that every node of @xmath18 is incident to exactly one edge in @xmath517 . * .. theorem.*[thm : perfect and secondary fan ] _ consider a subgroup @xmath41 as in [ introl ] .
let @xmath18 be a bi - partite graph which is obtained from @xmath41 as explained in section [ sec : dessin ] , and in definition [ graphs ] in particular .
let @xmath101 be a @xmath37-dimensional cone in the secondary fan of @xmath518 and let @xmath146 be the corresponding collection of @xmath37-element subsets of @xmath142 as in [ def : lsecondary fan ] and [ def : lsecondary polytope ] .
identify @xmath2 with the set of edges of @xmath18 . _
+ _ then , with the notation as in ( [ eq : quadruples ] ) , the set @xmath519 is a perfect matching on @xmath18 .
_ * proof . * consider a perfect surface @xmath520 as in [ perfect surface ] .
the map @xmath521 which is dual to the inclusion @xmath135 projects @xmath404 homeomorphically onto the plane @xmath136 .
the tiling of @xmath404 by unit squares gives a tiling of @xmath136 by parallelograms .
a unit square with edges @xmath133 and @xmath413 projects onto a parallelogram with edges @xmath190 and @xmath425 .
now let @xmath101 be a @xmath37-dimensional cone of the secondary fan ( see [ def : lsecondary fan ] ) and let @xmath501 be a white vertex of the tiling . by translating the origin of the secondary fan to @xmath501 one sees that there is exactly one parallelogram with vertex @xmath501 and with edges @xmath190 and @xmath425 such that @xmath522 , i.e. such that @xmath216 .
this shows that there is exactly one element of @xmath523 incident to @xmath501 , namely the black - white diagonal of this parallelogram .
a similar argument works for the black vertices . @xmath126 * .. corollary.*[black vola ] _ in the situation of theorem [ thm : perfect and secondary fan ] the number of black vertices and the number of white vertices are both equal to the number @xmath365 in ( [ eq : vola ] ) . _
* proof . *
the number of black ( resp .
white ) vertices of @xmath18 is equal to the number of edges in any perfect matching .
this holds in particular for the perfect matching @xmath523 corresponding to a @xmath37-dimensional cone @xmath101 in the secondary fan .
the result now follows from ( [ eq : pclc ] ) , ( [ eq : lc ] ) and ( [ eq : vola ] ) .
@xmath126 the following result was also derived in @xcite 5.1 . * .. corollary.*[genus ] _ the genus of the surface @xmath5 in [ twist ] is equal to the number of lattice points in the interior of the secondary polytope @xmath186 in ( [ eq : easy secondary ] ) . _
* proof . *
the number of @xmath20-cells in the tiling of @xmath5 is twice the number of @xmath37-cells . according to proposition [ count cells ]
the number of @xmath37-cells is @xmath524 .
the number of @xmath98-cells is , according to corollary [ black vola ] and formula ( [ eq : vola ] ) equal to @xmath525 .
the desired result now follows from corollary [ internal secondary ] and a calculation of the euler characteristic @xmath526 . @xmath126
* .. * [ traditional kasteleyn matrix ] in [ periodic models 1 ] we found the @xmath479-periodic bi - partite graph @xmath480 in the plane @xmath136 .
periodic bi - partite graphs in the plane are also known as _ periodic dimer models_. one of the main tools for studying periodic dimer models is the _
kasteleyn matrix_. it is usually defined as an adjacency matrix with extra factors @xmath527 for edges crossing the sides of a fundamental parallelogram and determined by visual inspection of a picture of the periodic dimer model ; see e.g. @xcite . the purpose of the extra factors is to keep track of the homology class of a closed path on the bi - partite graph @xmath18 in the torus @xmath498 ( see [ graphs ] ) .
we obtained @xmath528 as the projection of the black - white diagonals in the tiling of the perfect surface @xmath404 by unit squares .
the advantage of having @xmath404 embedded in @xmath45 is that these diagonals are actual vectors in @xmath39 and that points lying on paths formed by these diagonals are simply given by vector addition .
thus all edges become `` equally responsible '' for the periodicity .
an edge which is the diagonal in a square of type @xmath96 contributes @xmath472 .
thus we are led to a reformulation of the kasteleyn matrix of the dimer model @xmath529 in terms of the combinatorial data given as the list @xmath515 in equation ( [ eq : quadruples ] ) . as this same list
is used in [ twist ] for building the surface @xmath5 we call the same matrix also the bi - adjacency matrix of the dessin @xmath10 . * .. definition.*[def : kasteleyn matrix ] from the list @xmath530 in ( [ eq : quadruples ] ) and a function @xmath29 on the set of edges of the quiver @xmath1 we construct a matrix @xmath32 with rows indexed by the set of black vertices @xmath504 , columns indexed by the set of white vertices @xmath505 and with entries in the polynomial ring @xmath531 $ ] : @xmath532 we use for @xmath32 the name _ kasteleyn matrix of the dimer model @xmath529 _ as well as the name _ bi - adjacency matrix of the dessin @xmath10_. * .. example.*[dp3iv kasteleyn ] although the relation between the list @xmath515 and the matrix @xmath32 is so straightforward , that there seems no point in writing here the matrix @xmath32 for example [ correct dp3iv ] , we do present this @xmath32 here to compare it with results in the literature .
@xmath533 @xmath534\ ] ] ( with @xmath535 denoting @xmath536 ) . in view of the discussion in example [ dp3ivtwist ] this bi - adjacency matrix for model iv of @xmath231 should be somehow the same as the kasteleyn matrix for the singularity @xmath492 in @xcite equation ( 3.5 ) : @xmath537\,.\ ] ] indeed , by dividing the first , third and fifth column of @xmath32 by @xmath538 and subsequently setting @xmath539 , @xmath540 , @xmath541 , and appropriately setting @xmath542 one obtains a matrix which up to a permutation of rows and columns is @xmath543 .
this way of getting @xmath543 from @xmath32 is nothing but passing from homogeneous to inhomogeneous coordinates .
* .. definition.*[def : constellation](cf .
@xcite definition 1.1.1 . ) a _ @xmath47-constellation _
@xmath544 consists of a finite set @xmath2 and two permutations @xmath545 of @xmath2 such that the permutation group generated by @xmath545 acts transitively on @xmath2 .
* .. notations.*[cycle notations ] the cycle notation @xmath546 denotes the permutation @xmath547 such that @xmath548 for @xmath549 and @xmath550 . by a _
cycle _ of a permutation @xmath551 we mean an orbit of the group generated by @xmath551 .
we denote the set of cycles of @xmath551 by @xmath552 . in the product @xmath553 of two permutations @xmath551 and @xmath554 one first applies @xmath554 ; so @xmath555 . * .. definition.*[def : superpotential ] for our purposes we do not need the most general notion of a superpotential for a quiver .
the superpotentials we need are just notational reformulations of @xmath47-constellations .
the _ superpotential attached to a @xmath47-constellation _
@xmath544 is the following polynomial @xmath556 in non - commuting variables @xmath557 ( @xmath6 ) : @xmath558 where @xmath559 for a cyclic permutation @xmath560 .
* .. definition.*[def : dessin constellation ] the _ @xmath47-constellation @xmath544 associated with the dessin @xmath10 _ in [ final m ] is the following .
@xmath2 is the set of arrows of the quiver @xmath1 .
every @xmath6 is a path @xmath7 on @xmath5 .
every connected component of @xmath561 has an oriented boundary which can be viewed as a cyclic permutation of elements of @xmath2
. then @xmath562 ( resp .
@xmath563 ) is the composition of the cyclic permutations which are boundaries of connected components containing a black ( resp .
white ) point .
* .. remark.*[def : twist constellation ] in the same way one associates a @xmath47-constellation with the dessin @xmath564 in [ graphs ] .
it is obvious from the untwisting procedure in [ twist ] that , if @xmath544 is the @xmath47-constellation associated with the dessin @xmath10 , then @xmath565 is the @xmath47-constellation associated with the dessin @xmath564 . * .. * [ m from constellation ] it is obvious that the @xmath47-constellation @xmath544 associated with the dessin @xmath10 contains the complete instructions for building @xmath5 : for every cycle of @xmath562 and every cycle of @xmath563 take a ( convex planar ) polygon with sides labeled by the elements in the cycle in their cyclic order .
next glue these polygons by identifying sides with the same label .
the sides of these polygons correspond to arrows of the quiver @xmath1 and the vertices of these polygons correspond to vertices of @xmath1 . after the above procedure of glueing polygons along their sides
one must still identify points which correspond to the same vertex of @xmath1 .
the result is then @xmath5 with @xmath1 embedded in it . * .. * [ bi - adjacency constellation ] the @xmath47-constellation @xmath544 in [ m from constellation ] and the list @xmath515 in [ twist ] both are completely determined by and do completely determine the dessin @xmath10 .
one can , however , also describe the relation between @xmath544 and @xmath515 in a direct algebraic / combinatorial way . for the construction of @xmath544 from @xmath515 one first forms for @xmath513 , @xmath514 and @xmath66 the sets @xmath566 ( resp .
@xmath567 ) are neighbors if and only if @xmath568 , while @xmath569 are neighbors if and only if @xmath570 . in order to put a consistent orientation on these cyclic sets
we choose one of the two possible orientations of @xmath571 . for every @xmath513 we have @xmath572 or @xmath37 . in the latter case @xmath573
consists of two elements , say @xmath23 and @xmath574 , which are neighbors both in @xmath571 and in @xmath575 .
the orientation on @xmath571 then induces an orientation on @xmath575 such that @xmath23 is the successor of @xmath574 in @xmath575 if @xmath23 is the successor of @xmath574 in @xmath571 . in this way
the orientation on @xmath571 induces an orientation on every @xmath575 and every @xmath576 which has a non - empty intersection with @xmath571 .
next this induces an orientation on every @xmath577 which has a non - empty intersection with any of the already oriented sets @xmath575 or @xmath576 .
and so on .
it is because of the geometric background of @xmath515 ( i.e. the orientations shown in the right - hand picture in figure [ fig : untwist helix ] ) that we can indeed go on . in the end all sets @xmath575 and @xmath576 have an oriented cyclic structure and can be identified with cyclic permutations @xmath578 and @xmath579 .
the construction of the @xmath47-constellation @xmath544 is finished by setting @xmath580 the reason for inverting the white cycles is shown in figure [ fig : constellation ] ( 350,200)(-100,-20 ) ( 200,0 ) ( 150,100)(0,0 ) ( 0,0)(1,1)80 ( 0,0)(-1,1)80 ( 0,160)(-1,-1)80 ( 0,160)(1,-1)80 ( 0,0 ) ( 0,160 ) ( -80,80)(1,0)120 ( -80,80)(1,0)160 ( -12,175)@xmath500 ( -12,-25)@xmath501 ( 85,80)@xmath502 ( -93,80)@xmath503 ( 0,140)(30,20)[b ] ( 0,20)(30,20)[t ] ( 5,30)(-1,0)15 ( -5,130)(1,0)15 ( -5,85)@xmath562 ( -5,66)@xmath563 ( -5,115)@xmath578 ( -5,35)@xmath579 to go from @xmath544 to @xmath515 one views the cycles of the permutations @xmath562 , @xmath563 and @xmath581 as elements of the sets @xmath582 , @xmath583 @xmath584 , respectively , and then sets @xmath585 with @xmath586 * .. example.*[dp3ivconstellation ] the procedure of [ bi - adjacency constellation ] converts the list @xmath515 in [ correct dp3iv ] into the @xmath47-constellation @xmath544 with @xmath587 since in this example the cycle @xmath588 meets each of the above @xmath47-cycles the conversion went especially fast .
the corresponding superpotential @xmath556 is @xmath589
* .. * [ kasteleyn determinant ] corollary [ black vola ] implies that the bi - adjacency matrix @xmath590 of the dessin @xmath10 is a square matrix .
its determinant is by definition @xmath591 where @xmath554 runs over the set of all bijections @xmath592 and @xmath593 is defined as the sign of the permutation @xmath594 of @xmath504 for some reference bijection @xmath595 . of course
, if we want to actually write @xmath32 as a matrix , we must choose bijections between @xmath504 , @xmath505 and the set of numbers @xmath596 and that fixes @xmath597 .
changing the reference bijection multiplies @xmath598 by @xmath599 , but that is for our purpose unimportant .
next note that the @xmath34-entry of @xmath32 is non - zero if and only if there is an edge of @xmath18 connecting the nodes @xmath500 and @xmath501 .
so the only bijections @xmath554 that contribute to the determinant of @xmath32 are the _ perfect matchings _ ; see definition [ def : perfect matching ] .
thus we find @xmath600 where @xmath601 and @xmath602 for @xmath603 . *
.. definition.*[def : newton polytope ] the _ newton polytope _ of a laurent polynomial @xmath604\ ] ] is the polytope @xmath605 * .. theorem.*[thm : kasteleyn and secondary polygon ] _ the newton polytope of the determinant of the bi - adjacency matrix @xmath32 of the dessin @xmath10 is the same as the secondary polygon of @xmath41 ( see [ def : lsecondary polytope ] ) : @xmath606 the vertices of @xmath607 are the points @xmath608 ( see ( [ eq : lift perfect ] ) ) given by the perfect matchings @xmath523 with @xmath101 a @xmath37-dimensional cone in the secondary fan .
_ * proof . * if @xmath517 and @xmath609 are two perfect matchings then the edges in @xmath517 oriented black - to - white together with the edges in @xmath609 oriented white - to - black form a collection of closed loops on the torus @xmath498 in [ graphs ] .
the first homology group of this torus is @xmath38 .
therefore @xmath610 lies in @xmath41 .
this shows that @xmath611 lies in a @xmath37-dimensional plane in @xmath45 parallel to @xmath328 . from equation ( [ eq : lift perfect ] ) one sees that for every perfect matching @xmath517
@xmath612 i.e. the @xmath90-th coordinate of the vector @xmath613 equals the number of edges in the perfect matching @xmath517 which intersect the @xmath90-th zigzag loop ( cf .
[ zigzagquiver ] ) .
it follows from proposition [ count cells ] that the number of edges of the @xmath90-th zigzag loop is @xmath614 .
since consecutive edges in a zigzag loop can never intersect the same perfect matching we see that for every perfect matching @xmath517 and every @xmath90 @xmath615 fix @xmath90 and let @xmath101 and @xmath109 be the two @xmath37-dimensional cones in the secondary fan containing the half - line @xmath138
. then @xmath616 and @xmath617 .
since @xmath618 we find @xmath619 thus we see that for @xmath90 and the perfect matchings @xmath523 and @xmath620 equation ( [ eq : secondary bound 1 ] ) is in fact an equality .
fix a @xmath37-dimensional cone @xmath101 in the secondary fan .
it is bounded by two half - lines @xmath138 and @xmath621 .
the above argument now shows that equation ( [ eq : secondary bound 1 ] ) is in fact an equality for the perfect matching @xmath523 and @xmath90 and for @xmath523 and @xmath622 in place of @xmath90 .
thus the point @xmath608 is a vertex of @xmath623 .
this together with ( [ eq : secondary bound 1 ] ) proves : @xmath624 to complete the proof of ( [ eq : newton is secondary ] ) we note that equations ( [ eq : lc ] ) , ( [ eq : psic ] ) , ( [ eq : pclc ] ) and ( [ eq : lift perfect ] ) together with proposition [ count cells ] show @xmath625 .
@xmath126 * .. example.*[dp3 model ii ] according to @xcite 8 the quiver for model ii of @xmath231 ( i.e. case @xmath626 in figure [ fig : dp3 models ] ) affords two different superpotentials .
this is confirmed by our algorithm , which yields the following two bi - adjacency matrices @xmath627\ ] ] @xmath628\ ] ] the corresponding superpotentials are : @xmath629 @xmath630 after figuring out the rule for translating the edge labels one finds that the prepotentials in eqs .
( 8.1 ) and ( 8.2 ) of @xcite are @xmath631 and @xmath632 . for both superpotentials the surface @xmath5 has genus @xmath20 .
figures 18 and 19 in @xcite show pictures of the planar periodic bi - partite graph which is the lifting of @xmath18 to the simply connected cover of @xmath5 .
one can not expect the kasteleyn matrices @xmath633 and @xmath634 in @xcite eq .
( 8.3 ) to be the same as the bi - adjacency matrices @xmath635 and @xmath636 , respectively , because of the `` untwist '' . @xmath635 and @xmath636 are kasteleyn matrices of the dimer models before the untwisting ( see [ def : kasteleyn matrix ] ) . moreover eqs .
( 8.4 ) and ( 8.5 ) of @xcite show that the newton polygons of the determinants of the kasteleyn matrices @xmath633 and @xmath634 are different . on the contrary , for both @xmath635 and @xmath636 the determinant is a linear combination of the monomials @xmath637}\!,\,{\mathbf{u}}^{[2,2,1,1,1,1]}\!,\,{\mathbf{u}}^{[1,1,1,2,2,1 ] } \!,\,{\mathbf{u}}^{[2,0,0,2,2,2]}\!,\ , { \mathbf{u}}^{[3,1,0,1,1,2]}\!,\,{\mathbf{u}}^{[0,2,2,2,2,0]}\!,\,{\mathbf{u}}^{[1,3,2,1,1,0]}\,.\ ] ] so , in agreement with theorem [ thm : kasteleyn and secondary polygon ] , the newton polygons of @xmath638 and @xmath639 coincide with the polygon in figure [ fig : dp3 models ] case @xmath626 and [ explict secondary polygon ] . in spite of having the same newton polygon @xmath638 and @xmath639 are not equal ; indeed when all weights are @xmath20 they differ in the coefficient of @xmath640}$ ] .
* .. * [ intro adet ] in @xcite p.297 eq . ( 1.1 ) gelfand , kapranov and zelevinsky define for a set @xmath641 as in [ introgkz ] the _ principal @xmath0-determinant @xmath642_. the definition uses the laurent polynomial @xmath643 where @xmath644 for @xmath645 and where the coefficients @xmath275 are variables .
the name `` principal @xmath0-determinant '' refers to the fact ( @xcite p.298 prop .
1.1 ) that @xmath642 can be written as the determinant of some exact complex , i.e. @xmath646 for certain matrices @xmath647 .
another useful description of @xmath642 is given in @xcite p.299 prop .
1.2 : @xmath648 where the product runs over all faces @xmath649 of the primary polytope @xmath311 ( cf .
[ a secondary ] ) ; @xmath650 is a multiplicity and @xmath651 is the _
@xmath652-discriminant _ of the laurent polynomial @xmath653 .
the latter discriminant is a polynomial in the variables @xmath654 with @xmath655 . to define it ( see @xcite p.271 )
one needs the algebraic set @xmath656 which is the closure in @xmath657 of : @xmath658 then , by definition , @xmath659 if @xmath660 and @xmath661 so , @xmath642 gives the locus of the points @xmath662 for which at least one of the laurent polynomials @xmath663 has a critical point with critical value @xmath98 .
* .. * [ motivate conjecture ] theorems [ thm : secadet ] and [ thm : kasteleyn and secondary polygon ] in combination with formula ( [ eq : two secondary polytopes ] ) make one wonder whether for an appropriate choice of the weight @xmath28 the determinant of the bi - adjacency matrix @xmath32 is equal to the principal @xmath0-determinant @xmath642 , up to the simple transformation necessitated by ( [ eq : two secondary polytopes ] ) . in order to formulate this transformation
we must make the dependence on @xmath275 visible by writing @xmath664 and @xmath665 . * .. conjecture.*[conj : critical weight ] _ for every set @xmath0 as in [ introgkz ] and every dessin @xmath10 ( see [ final m ] ) constructed from @xmath0 by the algorithm in section [ sec : dessin ] the determinant of the bi - adjacency matrix with critical weight and the principal @xmath0-determinant satisfy : _
@xmath667 * .. remark.*[secvertices ] _ in support of the above conjecture we can point out that the coefficients of the monomials corresponding to the vertices of the secondary polygons are , up to sign , the same for the two sides of equation ( [ eq : crit det ] ) .
_ indeed , theorem 1.4 of @xcitep.302 gives the coefficient of the monomial in the principal @xmath0-determinant with corresponds to a vertex of the secondary polytope .
such a vertex corresponds to a maximal cone @xmath101 in the secondary fan . in the notations of definition [ def : lsecondary polytope ]
the formula in loc .
cit . for the coefficient of the monomial corresponding to @xmath101 reads : @xmath668 on the other hand , theorems [ thm : perfect and secondary fan ] and [ thm : kasteleyn and secondary polygon ] show that the same vertex corresponds to a perfect matching @xmath523 . from the role of perfect matchings in the computation of the determinant of the bi - adjacency matrix ( see [ kasteleyn determinant ] ) one now easily checks that the coefficient of the monomial in the determinant , which corresponds to @xmath101 , is ( possibly up to sign ) the same as ( [ eq : vertexcoeff ] ) . * .. example.*[dp1 kasteleyn ] for case @xmath669 $ ] in figure [ fig : models ] the algorithm in section [ sec : dessin ] yields in the following bi - adjacency matrix with critical weights : @xmath670\,.\ ] ] one easily computes @xmath671}+4{\mathbf{u}}^{[1,2,3,0]}+4{\mathbf{u}}^{[2,1,0,3 ] } -18{\mathbf{u}}^{[1,1,2,2]}-{\mathbf{u}}^{[2,2,1,1]}\ ] ] note that the exponents are the same as the coordinates of the vertices and the interior point of the secondary polygon in [ secondary fan dp1 ] . for the computation of the principal @xmath0-determinant we note that in this case ( see also example [ cubic roots ] ) @xmath672,\ , \left[\begin{array}{c}1\\ 3\end{array}\right],\ , \left[\begin{array}{c}1\\ 1\end{array}\right],\ , \left[\begin{array}{c}1\\ 2\end{array}\right]\right)\ ] ] and hence @xmath673 .
the primary polytope is shown in figure [ fig : secondaries ] .
the two boundary points of this primary polytope contribute factors @xmath674 and @xmath675 to the principal @xmath0-determinant ( cf . equation ( [ eq : adetdisc ] ) ) .
the contribution from the full primary polytope is the very classical discriminant of the cubic polynomial @xmath676 , which is ( e.g. @xcite p.405 ) @xmath677 thus we find @xmath678 } + 4{\mathbf{u}}^{[2,1,0,3]}+4{\mathbf{u}}^{[1,2,3,0 ] } -{\mathbf{u}}^{[1,1,2,2]}-18{\mathbf{u}}^{[2,2,1,1]}\,.\ ] ] _ this shows that conjecture [ conj : critical weight ] holds in this case .
* .. example.*[dp3 determinants ] when critical weights are used the two bi - adjacency matrices in example [ dp3 model ii ] become @xmath679\ ] ] @xmath680\ ] ] one easily computes @xmath681 and @xmath682 and finds that both are equal to @xmath683}\,+2{\mathbf{u}}^{[1,1,1,2,2,1]}\,+{\mathbf{u}}^{[1,3,2,1,1,0 ] } \,+{\mathbf{u}}^{[3,1,0,1,1,2]}\,- \\ & & \hspace{7em } \,-{\mathbf{u}}^{[2,0,0,2,2,2]}\,-{\mathbf{u}}^{[0,2,2,2,2,0]}\,-6{\mathbf{u}}^{[2,2,1,1,1,1]}.\end{aligned}\ ] ] this gives a remarkable contrast with the last line of example [ dp3 model ii ] and underlines the role of the critical weight in making conjecture [ conj : critical weight ] reasonable for every dessin obtained from @xmath0 . transforming the above determinant as in the left - hand side of equation ( [ eq : crit det ] ) yields @xmath684}\,+2{\mathbf{u}}^{[3,3,3,2,2,3]}\,+{\mathbf{u}}^{[3,1,2,3,3,4 ] } \,+{\mathbf{u}}^{[1,3,4,3,3,2]}\,-{\mathbf{u}}^{[2,4,4,2,2,2]}\,- \\ \nonumber & & \hspace{14em } \,-{\mathbf{u}}^{[4,2,2,2,2,4]}\,-6{\mathbf{u}}^{[2,2,3,3,3,3 ] } \\[1.5ex ] \label{eq : factor det } & & \hspace{-4em}=\ : u_1u_2u_3 ^ 2u_4 ^ 2u_5 ^ 2u_6 ^ 2\,(u_4u_5-u_1u_2 ) ( 4u_3u_4u_5u_6+u_2 ^ 2u_3 ^ 2 - 2u_1u_2u_3u_6+u_1 ^ 2u_6 ^ 2).\end{aligned}\ ] ] from the matrix @xmath626 in figure [ fig : dp3 models ] one easily finds @xmath0 and the corresponding polynomial @xmath685 the primary polytope @xmath311 is @xmath47-dimensional and has seven @xmath37-dimensional faces , six of which are triangles and one is a quadrangle .
its only integer points are its vertices and these correspond to the monomials of @xmath686 .
one can compute the discriminants for the polynomials supported by the faces of the primary polytope and multiply these to get the principal @xmath0-determinant @xmath642 as in equation ( [ eq : adetdisc ] ) .
the result is exactly as in ( [ eq : factor det ] ) , with the @xmath220-term factor coming from the primary polytope itself , the @xmath37-term factor coming from the quadrangle face and the other factors with multiplicities coming from the vertices .
_ this shows that conjecture [ conj : critical weight ] holds in this case . _ * .. further examples.*[further ] exactly as in example [ dp1 kasteleyn ] one can verify conjecture [ conj : critical weight ] for other `` four - nomials '' of degree @xmath687 ( see example [ labc ] ) by using the formulas in @xcite for discriminants of quartic and quintic polynomials .
the method of example [ dp3 determinants ] can be successfully employed to verify conjecture [ conj : critical weight ] for , for instance , cases @xmath688 , @xmath689 , @xmath490 in figure [ fig : dp3 models ] .
it also works for @xmath690 $ ] and @xmath691 $ ] , which are closely related to @xmath241 and @xmath692 , respectively .
bruijn , n.g . de , _ algebraic theory of penrose s non - periodic tilings of the plane i _ , proceedings of the koninklijke nederlandse akademie van wetenschappen , series a , vol 84 (= indagationes mathematicae , vol 43 ) , 1981 , 39 - 52 gelfand , i.m . ,
zelevinskii , m.m .
kapranov , _ hypergeometric functions and toral manifolds , _ functional analysis and its applications 23 ( 1989 ) , 94 - 106 + @xmath693correction to @xcite , funct .
analysis and its appl .
27 ( 1993 ) 295 stienstra , j. , _ gkz hypergeometric structures _ , in : _ arithmetic and geometry around hypergeometric functions _ , r - p .
holzapfel , a. m. uluda , m. yoshida ( eds . ) , progress in mathematics vol .
260 , birkhuser , basel , 2007 , p. 313 - 371 ; see also math.ag/0511351 | this paper presents some parallel developments in quiver / dimer models , hypergeometric systems and dessins denfants .
it demonstrates that the setting in which gelfand , kapranov and zelevinsky have formulated the theory of hypergeometric systems , provides also a natural setting for dimer models .
the fast inverse algorithm of @xcite and the untwisting procedure of @xcite are recasted in this more natural setting and then immediately produce from the quiver data the kasteleyn matrix for dimer models , which is best viewed as the bi - adjacency matrix for the untwisted model .
some perfect matchings in the dimer models are direct reformulations of the triangulations in gkz theory and the rule which maps triangulations to the vertices of the secondary polygon extends to a rule for mapping perfect matchings to lattice points in the secondary polygon . finally it is observed in many examples and then conjectured to hold in general , that the determinant of the kasteleyn matrix with suitable weights becomes after a simple transformation equal to the principal @xmath0-determinant in gkz theory .
illustrative examples are distributed throughout the text . |
one way to tackle the problem of fluid turbulence is to study its model with a drastically reduced degrees of freedom .
these models are made in mostly phenomenological ways to qualitatively have a few , selected aspects of turbulence .
celebrated models include the burgers equation @xcite and the shell models @xcite , which are certainly more amenable to analytical and numerical studies .
another family tree of these models stems from the constantin - lax - majda ( clm ) model @xcite .
we find a turbulent solution of this model family for the first time and discuss its relevance to two - dimensional navier - stokes turbulence .
the clm model was introduced to study the problem of putative finite - time blowup of the incompressible euler equation in three dimensions , that is , whether or not the vorticity becomes infinite in a finite time starting from a smooth initial condition @xcite .
the clm model is a one - dimensional partial differential equation which models the three - dimensional ( 3d ) inviscid vorticity equation .
the velocity gradient in the vortex - stretching term is modeled as the hilbert transform of the scalar vorticity to incorporate the biot - savart nonlocality .
but the model omits the advection term .
significantly , the model has an explicit analytic solution blowing up in a finite time .
the natural question was soon addressed : does the ( hypo- ) viscous effect , let us say adding the laplacian ( or various orders thereof ) of the model vorticity , prevent a solution from blowing up ?
unfortunately the answer was no @xcite ; the vorticity invariably blows up in finite time however large order of the laplacian is added .
this is in contrast to the existence of 3d navier - stokes ( ns ) solutions with the laplacian of order larger than @xmath0 @xcite .
this indicates that the vortex stretching in the clm equation is modeled somewhat excessively .
given that the viscosity is not enough to suppress the blowup , what if the omitted advection term is retained in the clm model ?
this was first considered by de gregorio @xcite and it is numerically shown that there exists a unique smooth solution to the clm equation with the advection term globally in time @xcite , implying that the advection term is able to prevent the blowup , although a rigorous proof is unavailable as yet .
subsequently , in order to study balance between the advection and stretching terms for the existence of solutions , the generalized constantin - lax - majda - de gregorio ( gclmg ) model , @xmath1 , was introduced in @xcite . here
@xmath2 is a parameter discussed below , @xmath3 denotes the model vorticity and the model velocity @xmath4 is expressed in terms of the vorticity as @xmath5 and @xmath6 which is the hilbert transform of @xmath7 . the parameters @xmath8 and @xmath9 correspond to the clm model and the clm model with advection term , respectively .
note also that , for general @xmath10 , the galilean invariance is lost .
mathematically , the short - time existence of solutions of the gclmg equation is proven for all @xmath10 @xcite , while it is conjectured that there exists some critical value , @xmath11 , above which a solution exists globally in time @xcite .
here is a twist : negative @xmath10 can be considered @xcite at the expense of the analogy of the gclmg model to the 3d vorticity equation .
instead , we gain the conservation law : for @xmath12 , it is shown easily that @xmath13 is conserved @xcite . for @xmath14 , the blowup of @xmath15
is proven rigorously @xcite , while for @xmath16 only numerical evidence for the blowup is available . with these mathematical facts about the gclmg model , we address the following natural question : does a viscous gclmg model work as a physical model of turbulence ? it is thus reasonable to consider the following forced viscous gclmg model : @xmath17 where @xmath18 is the model kinematic viscosity and @xmath19 represents an external force . for @xmath9 ,
albeit the similarity to the 3d vorticity equation , our numerical result shows that the model s energy concentrates on the forcing scale and the inertial range ( ir ) is not developed at all .
this indicates that the model is inadequate for ns turbulence in the ir .
the reason for this inadequacy can be that the model admits no conserved quantity such as the energy .
on the other hand , for @xmath20 , the ( model ) enstrophy becomes the conservative quantity of the inviscid gclmg model .
such quadratic conservation law is an essential element of ns turbulence as stressed , e.g. , in @xcite . in this respect , the gclmg model is akin to the 2d ns equation rather than the 3d vorticity equation .
therefore , our aim in this paper is to investigate whether eq.([gclmg ] ) with @xmath20 serves as a model of the enstrophy - cascade turbulence of the two - dimensional ( 2d ) incompressible ns equations . among features of the enstrophy - cascade turbulence @xcite , we focus on similarities and differences of the energy spectrum , in particular , the kraichnan s logarithmic correction @xcite to the kraichnan - leith - batchelor ( klb ) energy spectrum , @xmath21 @xcite , and of the structure function of the vorticity .
we also try to describe the gclmg turbulence from a dynamical system point of view .
we numerically simulate the gclmg equation ( [ gclmg ] ) for @xmath20 in a periodic interval of length @xmath22 with a standard dealiased spectral method by assuming null vorticity fourier mode of the zero wavenumber using the fourth order runge - kutta scheme and the same filtering of the round - off noise as @xcite . to achieve a statistically steady state ,
the large - scale forcing @xmath23 is set to be random : its fourier coefficient @xmath24 is non zero only for the forcing wavenumbers @xmath25 , whose real and imaginary parts are set to gaussian , delta - correlated - in - time and independent random variables with zero mean and variance , @xmath26 .
specifically , with keeping @xmath27 ( the average enstrophy - input rate is thus @xmath28 ) , we vary the kinematic viscosity as @xmath29 with @xmath30 .
the corresponding time step , @xmath31 , and the number of grid points , @xmath32 , are @xmath33 . _
top inset _ : a typical snapshot of the vorticity with pulse structures and the velocity of the randomly forced gclmg equation .
_ main figure _ : the time - averaged energy spectra , @xmath34 , of the forced gclmg equation for three values of the viscosity .
the dashed line is eq.([e ] ) for the inertial range . _
bottom inset _ : the corresponding time - averaged enstrophy fluxes , @xmath35 .
the plateau value is equal to the average enstrophy input rate . ] in the top inset of fig.[spec ] , we show an example of the vorticity and velocity from the middle viscosity case .
it is seen that the gclmg turbulent state is characterized by several pulses of the vorticity , which move randomly and sometimes merge and emerge .
the energy spectra , @xmath36 ( @xmath37 is the velocity fourier coefficient ) , for the three cases are plotted in fig.[spec ] .
it exhibits the ir and dissipation range . indeed , as shown in the bottom inset of fig.[spec ] , the enstrophy flux , @xmath38 , in the fourier space of the gclmg equation @xcite has a plateau which extends to larger wavenumbers as we diminish the kinematic viscosity .
the enstrophy flux is defined as @xmath39 $ ] , where @xmath40 is the vorticity fourier coefficient and @xmath41 denotes complex conjugate . here
we regard the plateau region as the ir of the gclmg turbulence . having obtained evidence of the enstrophy cascade in the forced gclmg turbulence , however , we observe that the energy spectrum @xmath34 in the ir has a visible correction to the klb law , @xmath42 , where @xmath43 is the average enstrophy dissipation rate .
this law is the simplest dimensional - argument result on the enstrophy - cascade spectrum in the ir @xcite .
moreover , the gclmg spectrum deviates also from the kraichnan s prediction with the logarithmic correction about the 2d ns enstrophy - cascade turbulence , @xmath44^{-1/3}$ ] @xcite , where the forcing wavenumber is @xmath45 here .
if we fit @xmath34 with a pure power law , @xmath46 , without the logarithmic correction , the exponent is close to @xmath47 .
nevertheless , we observe that dependence of @xmath34 on @xmath43 is consistent with the klb law from collapse of @xmath48 onto a single curve in the inertial range for various enstrophy dissipation rates ( figure not shown ) .
this indicates that the scaling behavior of the model is determined by the enstrophy cascade .
the scaling of the energy spectrum will be examined in terms of the vorticity structure function .
even order ( @xmath49 ) structure functions of the vorticity for the smallest viscosity case .
the vertical dashed line corresponds to @xmath50 .
the inertial range is here estimated as @xmath51 from the bottom inset of fig.[spec ] in which the plateau extends from @xmath52 .
_ inset _ : logarithmic local slopes of the structure functions ( @xmath53 from top to bottom ) for the three viscosity cases .
oscillations tend to be reduced as we increase the number of statistical samples ( the slope for @xmath54 is not shown because they are so large ) .
the vertical dashed lines correspond to the respective smallest scales in the ir . ]
now we move to the @xmath55-th order structure function of the vorticity , @xmath56 , where @xmath57 and @xmath58 denotes space - time average . in fig.[stfo ] , @xmath59 are shown in the log - log coordinates .
it appears that @xmath60 is close to @xmath61 , which is consistent with @xmath62 .
however the logarithmic local slope , @xmath63 / [ \ln ( r + \delta r ) - \ln r]$ ] ( @xmath64 being the grid size ) , does not converge to a unique constant value in the ir as we decrease @xmath18 .
one possibility is that the power - law exponent of @xmath60 in the ir depends on @xmath18 ( similar dependence was found for the velocity structure function of the 2d enstrophy - cascade turbulence @xcite ) .
another possibility is that the dominant part of @xmath60 is not a pure power law but with a correction . for the fourth order
, it is seen that @xmath65 is nearly constant in the ir .
now let us assume that the even - order structure function is a pure power law with exponent @xmath66 to the leading order , i.e. , @xmath67 .
then the relation @xmath68 , as indicated in fig.[stfo ] , implies that the vorticity can not be bounded ( see sec .
8.4 of ref.@xcite ) . with a given enstrophy input
, @xmath7 of the gclmg turbulence can be infinite for @xmath69 as we will see later .
this is in contrast to the velocity of the incompressible fluids .
conversely , if we consider that @xmath7 should be bounded , then @xmath70 is not pure power law but with a correction .
we now have two scenarios : ( i ) the pure power law , @xmath62 and @xmath71 , leading to infinite vorticity and ( ii ) the power law with a correction , @xmath72(non - power - law correction ) and @xmath73(non - power law part ) , with bounded vorticity .
we will later present evidence for ( ii ) while the infinite vorticity as @xmath69 is indicated simultaneously . for higher orders
, we observe that @xmath74 and @xmath75 have a peak in the dissipative range and that they are decreasing functions in the ir .
the peaks are due to the biggest pulse of the vorticity ( the rightmost one in the top inset of fig.[spec ] ) as expected .
the scale of the peak , @xmath76 , coincides with the typical width of the biggest pulse . about the decreasing behavior in the ir , it is unlikely to be an artefact caused by the peak since the decreasing region extends for smaller @xmath18 as seen in the bottom inset of fig.[stfo ] .
then do @xmath74 and @xmath75 have negative scaling exponents in the ir , as discussed for the 2d enstrophy - cascade turbulence @xcite ?
our answer is yes , but with non power - law corrections since the logarithmic slopes ( here shown only for @xmath74 ) are not satisfactorily flat .
the negative exponents are consistent with the singular behavior of the vorticity ( see @xcite ) .
we conclude that @xmath34 and @xmath59 of the gclmg turbulence in the ir are not described by a pure power law but with certain corrections and that @xmath59 s behavior indicates a singularity of the vorticity .
now we compare the above results of the gclmg turbulence with those obtained for the 2d ns enstrophy - cascade turbulence ( henceforth 2d turbulence ) .
we limit ourselves here to experimental and numerical results of the 2d turbulence having only the enstrophy - cascade ir as a statistically steady state ( not a decaying state ) .
concerning the energy spectrum , the difference from the klb @xmath21 scaling is more measurable in the gclmg turbulence than in the 2d turbulence .
the spectrum of the latter is occasionally fitted with the power law @xmath77 @xcite . regarding the vorticity structure functions of the 2d turbulence ,
the result obtained in the laboratory experiment @xcite showed that those of even orders up to @xmath78 in the ir have power - law exponents indistinguishable from zero .
this is consistent with the theoretical results beyond dimensional analysis of the 2d turbulence @xcite .
we note that in @xcite no indication of @xmath68 was observed , assuming that the structure function in the ir is a power - law function .
numerical results of the 2d turbulence showed that @xmath60 is logarithmic without a power law and that @xmath65 is a decreasing function @xcite , however .
switching to the velocity , we observe that both the even - order velocity structure functions and the even - order moments of the second - degree increments of the velocity @xcite of the gclmg turbulence are not power - law functions in the ir as indicated by their logarithmic local slopes without plateau .
this is in contrast with the 2d turbulence @xcite .
therefore , the gclmg turbulence obeys distinct statistics from the 2d turbulence despite the analogous enstrophy cascade .
however both have statistics with non power - law type corrections in common .
further details will be studied with a stationary solution of eq.([gclmg ] ) as follows .
surprisingly , the forced gclmg equation has a stable stationary solution whose energy spectrum is indistinguishable to that of the gclmg turbulence in the ir , as shown in fig.[statio ] .
we find it incidentally when we change the random forcing to a deterministic and stationary one , @xmath79 , in order to study dependence of the statistics on the large - scale forcing .
thus it demonstrates the striking forcing dependence .
we here fix the forcing amplitude @xmath80 and vary @xmath18 as @xmath81 .
we use the same spectral method and time stepping as before with the corresponding time step @xmath82 and the number of grid points @xmath83 .
the stationary solution has one vorticity pulse as depicted in the right inset of fig.[statio ] , which is likely to converge to a singular function ( infinite vorticity ) as @xmath69 ( analogously nonsmooth vorticity has been observed in the 2d stationary kolmogorov flow for small @xmath18 @xcite ) .
its peak value and width are numerically found to scale with @xmath84 and @xmath85 , leading to the enstrophy dissipation rate independent on @xmath18 .
the energy spectra of the stationary solutions to the gclmg equation with the deterministic forcing for various @xmath18 s .
the squares represent data ( multiplied by @xmath86 ) of the randomly forced case with @xmath87 shown in fig.[spec ] .
_ left inset _ : the logarithmic local slopes of the energy spectra and eq.([a ] ) as the dashed line .
_ right inset _ : the vorticity pulses of the stationary solutions . ]
the identity of the energy spectrum of the stationary solution to that of the gclmg turbulence in the ir implies that the functional form of the turbulent @xmath34 can be studied with the stationary solution . in the left inset of fig.[statio ] , we plot the plateau - less logarithmic local slope of @xmath34 , demonstrating no pure power law in the ir .
we then make the following ansatz in the ir , @xmath88^\delta , \label{a}\ ] ] although small discrepancy among @xmath18 s are present ( but not visible in the bottom inset of fig.[statio ] ) .
for @xmath89 , integration of eq.([a ] ) leads to the expression of the energy spectrum , @xmath90^{\theta}\right\ } , \label{e}\ ] ] where @xmath91 and @xmath92 . here
we fix the first parameter as @xmath93 in view of the enstrophy cascade and our previous observation of @xmath34 s dependence on the enstrophy dissipation rate .
the rest of the parameters are estimated as @xmath94 and @xmath95 by fitting eq.([a ] ) in the range @xmath96 , resulting in @xmath97 and @xmath98 .
so far eqs.([a]-[e ] ) are empirical ( but @xmath99 case can be obtained with the incomplete self - similarity @xcite ) .
the log - corrected spectrum of the form @xmath100 , corresponding to @xmath101 and @xmath102 , does not yield a better fit however we adjust @xmath103 , implying that this type of the correction is not suitable for the gclmg turbulence .
self - similarity of the phase - space orbits of @xmath104 $ ] of the randomly forced gclmg turbulence ( @xmath105 case ) for @xmath106 ( top ) and @xmath107 ( bottom ) as @xmath108 variables from two angles of view .
the point corresponds to the stationary solution with the same enstrophy and viscosity.,title="fig : " ] self - similarity of the phase - space orbits of @xmath104 $ ] of the randomly forced gclmg turbulence ( @xmath105 case ) for @xmath106 ( top ) and @xmath107 ( bottom ) as @xmath108 variables from two angles of view .
the point corresponds to the stationary solution with the same enstrophy and viscosity.,title="fig : " ] self - similarity of the phase - space orbits of @xmath104 $ ] of the randomly forced gclmg turbulence ( @xmath105 case ) for @xmath106 ( top ) and @xmath107 ( bottom ) as @xmath108 variables from two angles of view .
the point corresponds to the stationary solution with the same enstrophy and viscosity.,title="fig : " ] self - similarity of the phase - space orbits of @xmath104 $ ] of the randomly forced gclmg turbulence ( @xmath105 case ) for @xmath106 ( top ) and @xmath107 ( bottom ) as @xmath108 variables from two angles of view .
the point corresponds to the stationary solution with the same enstrophy and viscosity.,title="fig : " ] to see further relation of the stationary solution to the gclmg turbulence from a dynamical system point of view , we plot in fig.[att ] an orbit of the randomly forced case in the three dimensional subspace of the phase space , which is defined as @xmath109 / \omega(k_1 ) - 1,\ , { \rm im}[\hat{\omega}(k_2 , t ) ] / \omega(k_2 ) - 1,\ , { \rm im}[\hat{\omega}(k_3 , t ) ] / \omega(k_3 ) - 1 ) $ ] . here @xmath104 $ ] and @xmath110 are the imaginary part of the vorticity fourier coefficient of the randomly forced gclmg equation and the stationary solution , respectively .
we take wavenumber triplet @xmath111 as @xmath112 and @xmath113 which are in the ir except for @xmath114 .
hence we look at the orbit scaled with the stationary solution from the two ranges of scales .
we observe that the orbit is within a thin surface with the stationary solution located on one edge of this attracting set .
significantly , the orbits in the two scale ranges are similar . with this self - similarity of the orbit , modeling of the gclmg turbulence with a few degrees of freedom
is conceivable .
we numerically studied the gclmg equation eq.([gclmg ] ) with @xmath20 with a large - scale forcing .
the random forcing generates the turbulent state analogous to the 2d ns enstrophy - cascade turbulence due to existence of the inviscid quadratic invariant , the enstrophy .
the deterministic forcing yields the stable stationary solution which is spectrum - wise relevant to the gclmg turbulence .
the statistical laws of the gclmg turbulence , such as the measurable corrections to the klb law , are quantitatively different from the 2d turbulence
. however , qualitatively , statistics not characterized by simple power laws are common .
therefore we expect that the simpler gclmg model provides insight in theoretical study of these subtle statistics , which are not captured by dimensional analysis . currently , we failed to substantiate our expectation although the stationary solution should facilitate analytical study .
another remarkable aspect of the gclmg solutions is the strong indication of infinite vorticity with a finite enstrophy dissipation rate as @xmath69 , possibly an inheritance from the clm model .
this implies that the gclmg model is an interesting testing ground for investigating relation between singularity and turbulence statistics .
however , we note that the finite - limit of the enstrophy dissipation rate is not shared with the 2d enstrophy - cascade turbulence ( see , e.g.,@xcite ) . for different negative @xmath10 s in eq.([gclmg ] ) , we found that the similar results to the @xmath16 case holds , which will be reported elsewhere .
we acknowledge delightful discussions with koji ohkitani , hisashi okamoto and shin - ichi sasa and the support by grants - in - aid for scientific research kakenhi ( b ) no .
26287023 from jsps . | as a minimal mathematical model generating cascade analogous to that of the navier - stokes turbulence in the inertial range , we propose a one - dimensional partial - differential - equation model that conserves the integral of the squared vorticity analogue ( enstrophy ) in the inviscid case . with a large - scale random forcing and small viscosity ,
we find numerically that the model exhibits the enstrophy cascade , the broad energy spectrum with a sizable correction to the dimensional - analysis prediction , peculiar intermittency and self - similarity in the dynamical system structure . |
quantum chromodynamics ( qcd ) owes its confining property to non - perturbative structures of its groundstate which are , after many years of research , not yet uniquely identified .
( for a recent overview of competing pictures see e.g. @xcite ) . in a more pragmatic attitude , the stochastic vacuum model ( svm ) @xcite describes the vacuum structure in terms of a few vacuum correlation functions and points out which are related to confinement . that only few correlators should characterize
the vacuum is the consequence of a truncation of the hierarchy of correlation functions , essentially due to a postulate made within the svm that the gauge fields are gaussian random processes in euclidean space - time . in this framework ,
the correlators enter , and relate to each other , various phenomenological predictions . particularly interesting
( exhausting , in the case of pure gluodynamics ) is the gauge invariant gluon field strength correlator in euclidean space ( and its continuation to minkowski space ) . according to the svm , part of the euclidean field strength correlator at zero temperature
is related to the confinement property of the qcd ground state @xcite . for heavy quarkonium states ,
the full correlator describes the effect of the gluon condensate on the level splitting in detail ( see ref . @xcite and references therein ) .
wick - rotated to minkowski space , the correlator serves to parametrize high energy hadron - hadron scattering ( see e.g. ref .
@xcite ) .
regardless of how this truncation exactly works , these correlators incorporate the unknown non - perturbative mechanisms and can be considered as an implicit parametrization of the important gluonic vacuum field modes .
if they were known , this information could be readily used to constrain the parameters of more explicit models of vacuum structure .
the central role , both theoretically and phenomenologically , strongly suggests to extract the field strength correlator ( and other vacuum correlators ) from a first principle lattice simulation .
it is interesting to notice that , although the svm approach has been proposed more than a decade ago by dosch and simonov @xcite , on the lattice the field strength correlator has only been studied by the pisa group . apart from earlier results for @xmath0 @xcite and @xmath2 @xcite pure gauge theory , there are now precise results available , from an analysis using cooling , over an interval of distances 0.1 ... 1.0 fm for pure @xmath2 gluodynamics @xcite , for qcd with dynamical ( kogut - susskind ) fermions @xcite ( four flavors , for two values of mass ) over an interval 0.4 ... 1.0 fm , and , over the same interval of distances , for pure @xmath2 yang - mills theory very near the deconfining temperature @xcite . lattice data for @xmath3 obtained and fitted in refs . @xcite
give strong hints in favor of dominance of locally selfdual or antiselfdual field strengths .
these data have been analyzed @xcite from the point of view of the random instanton liquid model @xcite . here
the following picture arises : a rather dilute liquid of instantons ( with parameters which , in the case of full qcd , happen to be roughly consistent with parameters obtained from phenomenological analyses ) can explain the field strength correlator , including the part which is believed to be responsible for confinement .
the relevance of this statement for confinement is far from clear .
absolute confinement is , strictly speaking , a property of pure yang - mills theory alone . at this place
let us mention that a much higher instanton density is extracted from the field strength correlator in the case of pure yang - mills theory than for full qcd @xcite . on the other hand
, according to a widespread opinion a random instanton liquid , however dense it might be , does not lead to an asymptotic string tension .
this is obvious for calculations of the direct effect of the ( unperturbed ) instanton liquid on the static @xmath4-@xmath5 potential via the wilson loop .
while the instanton potential rises linearly at intermediate distance , this does not correspond to the known string tension @xcite if the phenomenological density and size are assumed to hold .
let us note , however , that neither instanton suppression by external charges or fields nor instanton interactions ( see _ e.g. _ @xcite ) have been taken into account in these estimates .
it has even been questioned @xcite whether the available lattice evidence for confinement is conclusive at all . in short , in spite of the possibility of an instanton interpretation of the correlator , the role of instantons for confinement is still controversial .
the answer depends on details of the modeling of the instanton liquid , in particular on the density and the shape of size distribution ( see e.g. refs .
@xcite ) . there is an indirect route ( not completely worked out yet ) leading from the instanton liquid to confinement .
it has been demonstrated that under certain circumstances an instanton liquid induces the condensation ( percolation ) of magnetic monopoles ( see ref.@xcite and earlier references therein ) which should confine via a dual meissner effect .
a corresponding effective theory of a dual abelian gauge field , interacting minimally with the ( condensed ) monopole field ( dual abelian higgs model @xcite ) is characterized by the mass of the dual `` photon '' and a finite higgs correlation length . both quantities together determine @xcite the gauge invariant non - abelian field strength correlator which we are interested in .
therefore , in a more fundamental way , the two structure functions of the correlator at @xmath3 are inherent to an infrared effective dual formulation of qcd ( see also @xcite ) . similarly to @xmath3 the field strength correlator at finite temperature encodes information on generic non - perturbative structures of non - abelian gauge field histories dominating the qcd partition function . in a similar spirit as of ref .
@xcite , the correlators obtained near the deconfinement transition in pure @xmath2 gauge theory @xcite have been analyzed from the caloron ( finite @xmath6 instanton ) point of view @xcite .
such a description turns out to be possible only below the deconfinement temperature @xmath7 . as any measurement involving the field strength ,
such lattice calculations are far from trivial .
the cooling method has been used in most of the more recent studies . to our opinion ,
the implications of this method for the intended measurements have not been discussed thoroughly enough so far . in this paper
, we are going to apply a different technique to improve the signal for extracting the field strength correlator from lattice gauge field ensembles .
our method is based on a ( classically ) perfect action and uses a renormalization group motivated smoothing procedure @xcite .
we shall give a presentation of first results obtained in this manner for finite temperature @xmath0 pure gauge theory .
in contrast to @xmath3 , a similar splitting of correlation lengths related to the different structure functions is obtained with both techniques .
some possible differences should be closer examined whether they are due to the other gauge group @xmath0 used here , compared to @xmath2 analyzed with cooling .
if not then , cooling would have a systematic effect concerning the relative strength of the structure functions @xmath8 and @xmath9 at high temperature .
the paper is organized as follows . in sec .
ii we give an introduction to the concept of field strength correlators for zero and finite temperature ; the correlators and their suitable decompositions into the structure functions are defined there . in sec .
iii we briefly discuss the potential problems of the cooling method and recall in short the fixed point action renormalization group ( rg ) smoothing technique . in sec .
iv we describe our results , and conclude with a discussion in sec . v.
the two - point field strength correlator is a non - local generalization of the gluon condensate , involving the field strength tensor at separate points : [ eq : basic_def ] d_(x ) = g^2 0| \ { g_(0 ) s(0,x ) g_(x ) s^(0,x ) } strength , with @xmath10 being the generators of the algebra of the color @xmath11 group in the fundamental representation .
@xmath12 is the ( path ordered ) schwinger line ( parallel transporter ) [ eq : schwinger_line ] s(x_1,x_2 ) ( i g _ x_1^x_2 dz^a_(z ) ) , with @xmath13 .
it is necessary to include the non - abelian schwinger line phase factor in order to ensure gauge invariance .
usually the straight line path from @xmath14 to @xmath15 is chosen to evaluate @xmath12 .
the correlator describes the coherence of the gauge field ( modulo gauge transformations ) over some coherence length .
this notion replaces the idea of `` domains '' originally imagined to form the gluon condensate .
completely contracted , it is the local gluon condensate defined as [ eq : gluon_cond ] g_2 : g^a _ g^a _ : , ( _ s = g^2 4 ) , which is obtained in the limit @xmath16 from the regulated correlator . at @xmath3 ,
euclidean @xmath17 invariance admits the decomposition ( for @xmath18 ) [ eq : decomp_t0 ] + & & + ( x_x _ _ - x_x _ _ + x_x _ _ - x_x _ _ ) _ 1(x ) x^2 , where @xmath8 and @xmath9 are two invariant functions ( of euclidean modulus @xmath19 ) containing information on the vacuum . the tensor structure related to @xmath9
does not contribute to confinement , while @xmath20 is possible only in non - abelian theories or in abelian theories with monopoles .
it is convenient , in particular on the lattice , to define two other vacuum structure functions @xmath21 and a @xmath22 , according to whether the plaquettes defining the field strength component in question have one direction parallel to the 4-vector between the two points of observation .
these functions are composed of @xmath8 and @xmath9 as follows [ eq : par_perp ] d _ & & d + d_1 + x^2 _ 1 x^2 , + d _ & & d + d_1 .
the two basic vacuum structure functions at @xmath3 have a non - perturbative and a perturbative part according to the notation [ eq : nonpert_pert ] d(x ) & = & d^np(x ) + a_0 |x|^4 e^-|x|/_0 , + d_1(x ) & = & d_1^np(x ) + a_1 |x|^4 e^-|x|/_1 , since the first structure function is closely related to confinement , the non - perturbative parts of the two functions are suggestively written as [ eq : non_pert ] d^np(x ) & = & ^2 6 g_2 ( x ) , + d_1^np(x ) & = & ^2 6 ( 1- ) g_2 _ 1(x ) , where @xmath23 has to be identified with the gluon condensate ( [ eq : gluon_cond ] ) and [ eq : normaliz ] ( 0 ) = _ 1(0 ) = 1 .
the so - called non - abelianicity @xmath24 represents a certain fraction of the gluon condensate related to confinement .
so far the functions @xmath25 and @xmath26 mostly have been fitted as exponential functions @xcite of the modulus @xmath19 [ eq : shape ] ( x ) = e^-|x|/_0 _ 1(x ) = e^-|x|/_1 , although more physically motivated forms are available which fit the data as well @xcite . in the svm
the string tension can be approximately related to the correlation length @xmath27 ( coherence length of @xmath8 ) and the gluon condensate @xcite , @xmath28 \\ [ -.25 cm ] & = & \frac{\pi^3}{12}~\kappa~g_2 \int dx^2 \tilde{\cal d}(x ) = \frac{\pi^3}{6}~\kappa~g_2~\lambda_0 ^ 2 \ , .
\nonumber\end{aligned}\ ] ] in the perturbative regime ( at short distance ) both @xmath8 and @xmath9 behave as @xmath29 .
one - gluon exchange contributes only to the latter @xcite giving @xmath30 .
one - loop and higher order perturbative contributions to the string tension integral ( [ eq : sigma ] ) are cancelled by higher correlator contributions @xcite .
purely selfdual or antiselfdual non - perturbative background fields contribute only to @xmath8 ( see ref .
@xcite ) . in this case
, it is the overlap of different selfdual or antiselfdual configurations ( higher order contributions in an expansion in the density @xcite ) which generates the non - perturbative part of @xmath9 .
the observation that @xmath8 @xmath31 @xmath9 , first made for @xmath3 lattices , supports the idea that strong fields are dominantly selfdual or antiselfdual .
it was only for @xmath3 that the newer data have been analyzed by the authors themselves @xcite .
as short distances became observable at higher @xmath32 it has been confirmed that the functions @xmath8 and @xmath9 contain a perturbative part which could be subtracted by a fit like ( [ eq : nonpert_pert ] ) . for the case of
quenched @xmath2 gauge theory the two correlation lengths have been found equal , @xmath33 fm .
gauge theory , a first analysis @xcite without cooling has found @xmath34 fm . ] the data for full qcd show that the correlation length @xmath27 becomes smaller with increasing quark mass towards the quenched case .
for the chiral limit an upper limit for the correlation length can be estimated as @xmath35 fm .
the gluon condensate @xmath23 as extracted from the correlator according to ( [ eq : nonpert_pert],[eq : non_pert ] ) is much larger in pure yang - mills theory than any phenomenological value . in ref .
@xcite the field strength correlator data at intermediate distance have been used to extract the two parameters characterizing the random instanton liquid model @xcite which gives a good description of most of the hadronic correlators at zero temperature . of the correlator @xmath36 can not be described .
the inclusion of the interaction with the vacuum medium into the `` classical '' euclidean field equation results in a modified instanton profile far from the center @xcite . for a recent interesting attempt to find a modified instanton solution and to use it for the correlator see ref @xcite . ]
the size distribution of this model is approximated as @xmath37 , so density @xmath38 and size @xmath39 are the only parameters .
the investigation in ref .
@xcite was carefully dealing with the path dependence of the schwinger factor in the instanton background . for a straight line path ,
a reasonable description over the measured range of distances has been achieved with parameters roughly compatible with phenomenological estimates where this is possible : density @xmath40 and size @xmath41 fm for full qcd ( with four flavors ) , to compare with @xmath42 and @xmath43 fm in pure @xmath2 yang - mills theory .
second order terms in @xmath38 give subleading contributions to @xmath8 and are the leading ones in @xmath9 .
as long as deviations from selfduality or antiselfduality are entirely due to @xmath44 , @xmath45 and @xmath46 overlap , this analysis has led to the conclusion that @xmath47 is characteristic for the instanton liquid at @xmath3 .
although this was not the case in the original fit given by the authors , it has been shown that the quenched data @xcite are compatible with a small negative @xmath9 as well . with the second order terms taken into account for pure yang - mills theory
, the best fit leads to a change towards smaller @xmath48 and @xmath49 fm .
the smallness of the correction _ a posteriori _ justifies the dilute gas approach to pure yang - mills theory and even more so for full qcd . as a rule one can summarize that instanton radii estimated in this way are @xmath50 % to @xmath51 % bigger than the ( exponential ) correlation length of @xmath8 . at finite temperature ,
the @xmath17 space - time symmetry is broken to the spatial @xmath52 symmetry and the number of independent correlators is enlarged@xcite .
all of them separately depend on @xmath53 and @xmath54 .
we consider @xmath55 where @xmath56 is the electric field , and @xmath57 where @xmath58 is the magnetic field .
let us assume that @xmath59 and define longitudinal and transversal correlators for electric and magnetic fields @xmath60 in addition there is an electric - magnetic correlator which will not be considered here . as a first step , in ref .
@xcite an exponential plus perturbative fit in terms of @xmath61 and @xmath62 has been tried to describe the finite @xmath6 quenched @xmath2 correlator measured in @xcite with cooling ( which is known at distances @xmath63 fm ) . for the magnetic correlators , immediately below the deconfining transition , at a temperature @xmath64 , the two structure functions have clearly different correlation lengths , @xmath65 fm and @xmath66 fm .
the first one slowly increases across the transition , while the second goes through a maximum there .
for the two electric structure functions at the temperature @xmath64 , different correlation lengths @xmath67 fm and @xmath68 fm have been found , too .
these fits show a tendency for both electric correlation lengths to drop suddenly at the transition .
unfortunately , there is no cooling data for pure @xmath0 at finite temperature that would allow a more direct comparison .
a description using the harrington - shepard caloron solution , i.e. , the periodic instanton @xcite , is the main objective in ref .
@xcite . below the deconfinement temperature , where @xmath69 and @xmath70 has been found and where @xmath71 , a satisfactory description of magnetic and electric correlators was possible using these selfdual or antiselfdual configurations while the small differences could be attributed to overlapping two - caloron ( or caloron - anticaloron ) configurations .
the ratio @xmath72 strongly increases above the transition . after describing our smoothing method in the next section
, we will discuss the correlation lengths for the different finite temperature structure functions according to exponential anstze similar to what has been done in ref .
@xcite for the cooled @xmath2 data , making combined fits to electric and magnetic data separately , in terms of @xmath73 and @xmath74 and in terms of @xmath75 and @xmath76 , respectively .
we will also discuss a fit which directly applies to the _ raw data _ for @xmath77 and @xmath78 simultaneously , as well as for @xmath79 and @xmath80 .
in this case we will separate the data into @xmath8 and @xmath9 afterwards , using the obtained form of the fit .
it should be emphasized already here that , due to our smoothing method , these functions are obtained essentially free of a perturbative contribution .
the pisa group @xcite has used the cooling method in order to measure the field strength correlators at @xmath3 , both without external sources ( in vacuum ) and in the presence of an external wilson loop @xcite , and at temperature near @xmath7 . for the measurement ,
similar to the search for instantons , it turned out necessary to eliminate ultraviolet fluctuations .
it has been argued that , due to the diffusive nature of cooling , the correlator to be measured at a distance @xmath81 ( @xmath82 lattice spacing ) would not be affected before a number @xmath83 of cooling iterations would have been performed . practically , one measures the expectation values defining @xmath84 and @xmath85 as a function of @xmath86 .
the actual correlator is then defined as the plateau value _ vs. _ @xmath86 .
although it has not been presented in detail , experience seems to justify the expectation that this plateau is reached safely before @xmath87 happens to fall within the diffusion radius @xmath88 . in another context
it has been shown that this amount of cooling amounts to a renormalization of coupling , which is expected to show up in the perturbative signal contributing to the correlator .
if cooling is technically necessary to improve the signal - to - noise ratio , one could be concerned to what extent it distorts the non - perturbative modes that existed in the true vacuum , i.e. , to what extent the correlation lengths and the strength of the correlators are affected by the procedure itself . since the ( exponential ) correlation length of the correlator @xmath8 is even smaller than the phenomenological instanton size it is useful to recall what effect cooling has in the attempt to fix instanton size and density from analyzing the topological density of a configuration . for this purpose ,
the cooling technique nowadays is superseded by adopting two , essentially opposite extreme ways to control that effect .
the first prescription requires to observe the average size as function of @xmath86 , extrapolating it backward to zero cooling steps @xcite in order to obtain the would - be instanton size in vacuum .
the other uses a global stopping criterion @xcite requiring cooling to be stopped when the density of instantons becomes equal to the ( much better known ! ) topological susceptibility .
this second method implicitly adopts a non - interacting instanton gas picture , which differs principally from the instanton liquid .
it is recognized by now that one has to control cooling since instantons , once they become discernible by cooling or smearing , are steadily growing , becoming more and more classical , while small size instanton - antiinstanton pairs annihilate .
improved cooling @xcite aims to stabilize instantons with @xmath89 , still suffering from the latter problem . presently , there is no method agreed upon , such that the instanton size can not be uniquely defined on the lattice @xcite .
each of the methods seems to have a smoothing radius of the lattice fields ( depending on @xmath86 ) which should be well below the distances at which one measures the correlator . during the last two years
a method has been advocated @xcite that avoids propagating , iterative procedures to suppress uv fluctuations .
alternatively , it uses first blocking and subsequent inverse blocking with respect to a perfect action in order to accomplish the necessary smoothing of equilibrium configurations , while it preserves their long range structure beyond the smoothing scale .
so far , this method has been used exclusively to analyze the instanton @xcite and monopole structure @xcite of monte carlo generated configurations . as for monopoles , smoothing wipes out small monopole loops which are irrelevant for confinement ( similar to blockspin renormalizations directly applied to the magnetic currents @xcite ) . to carry out a theoretically and technically consistent analysis
, one has to implement a fixed point ( fp ) action .
the simplified fixed - point action @xcite for @xmath0 has been improved recently , and a suitably truncated version @xcite has been obtained .
it is also parametrized in terms of only two types of wilson loops , plaquettes ( type @xmath90 ) @xmath91 and tilted @xmath92-dimensional @xmath93-link loops ( type @xmath94 ) of the form @xmath95 and contains several powers of the linear action terms corresponding to each loop ( of both types ) that can be drawn on the lattice @xmath96 the parameters of this action as used in this work are reproduced in table i. a comparison reported in detail in @xcite shows that the new parametrization is near to perfect . with this action at hand
one can now proceed to perform renormalization group transformations on the lattice .
suppose we have a fine lattice of links @xmath97 and a coarse lattice of links @xmath98 covering the same physical volume .
being a classically perfect action , @xmath99 can be evaluated on both lattices and must satisfy the following condition @xmath100 configuration by configuration over a representative ensemble of equilibrium gauge field configurations @xmath98 , each in one - to - one correspondence to a fine configuration @xmath97 saturating the lower bound on @xmath101 provided by the right hand side .
@xmath102 is a certain non - negative functional related to the blockspin transformation ( see @xcite ) .
( [ eq : equality ] ) can be read as the saddle point equation for an integral over fine configurations defining the blocked action @xmath103 . neglecting the one - loop corrections
is justified in the classical limit for @xmath104 . the parameter @xmath105 (
not to be confused with the non - abelianicity factor mentioned above ) has been fixed in the context of the adopted blocking scheme by an optimization of the locality properties of the quadratic part of the action .
blocking means a mapping from the link fields @xmath97 on the fine lattice to link variables @xmath98 which describe the same configuration on the next coarser length scale .
this mapping is accomplished by the following construction : @xmath106 the coarse link variable @xmath107 is obtained by normalization to a @xmath0 group element @xmath108 .
the blocking parameters @xmath109 and @xmath110 have been optimized earlier in ref .
@xcite under rather general conditions .
inverse blocking is a mapping @xmath111 where the new link variables @xmath112 are implicitly defined as the set of links providing , for the configuration encoded in @xmath98 , the `` smoothest interpolation '' on the next finer grained lattice .
practically , the link field @xmath113 is found by minimizing the extended action on the left hand side of eq .
( [ eq : equality ] ) .
this is not a truly classical field on the lattice but it is smoother than any quantum field @xmath97 that could give rise to the coarse level field @xmath98 . technically speaking
, @xmath114 plays here a role similar to a lagrange multiplier defining the strength of the constraint imposed by the coarse grained configuration .
the @xmath0 pure gauge field configurations to be analyzed below have been generated on a @xmath115 lattice at various @xmath32 with the ( classical ) perfect action . in ref .
@xcite we have localized the deconfinement phase transition at @xmath116 .
corresponding to the second order character of that phase transition , the intersection of binder cumulants of polyakov loops for different @xmath117 lattice volumes had been used for this purpose .
although the simulation was done on the fine lattice , the binder cumulants built from @xmath98 lattice polyakov loops give the same @xmath118 . while the polyakov line correlators become critical at the transition , the various field strength correlation lengths remain finite across the transition . here
we demonstrate how we can get immediate access to the non - perturbative part of the field strength correlators applying the rg smoothing method . with some reservations ,
the results can be compared with cooling analyses ( unfortunately available only for @xmath2 ) .
the field strength components @xmath119 are defined as the average of the antihermitean part @xmath120 of the four equal - oriented `` clover leaf '' plaquettes in the @xmath121 plane which are open at the lattice site @xmath122 : g a^2 g_(x)= ( u_x,,+u_x,,-+u_x,-,-+u_x,- , - ) .
[ eq : gemunu ] for the purpose of the following study we take advantage of the fact that in the result of smoothing , similar to cooling , the renormalization constant becomes @xmath123 for the gluon fields .
we have measured only equal - time correlators , _
i.e. _ with a timelike argument @xmath124 . for lattice sites
@xmath125 separated not along one of the spacelike axes , the choice of the schwinger line is not unique . in that case
averaging of expression ( [ eq : basic_def ] ) has been performed over a random selection of ten shortest paths connecting @xmath14 and @xmath15 , each path defining an independent transporter @xmath12 . naturally , for @xmath77 , @xmath79 , @xmath78 , and @xmath80 , the field components @xmath126 , @xmath127 , @xmath128 and @xmath129 parallel or perpendicular to the 3-vector @xmath130 have been appropriately projected out of the field strength tensor @xmath131 .
the raw data for @xmath132 and @xmath133 ( confinement ) , for @xmath134 ( just above deconfinement ) and three temperatures in the deconfined phase ( @xmath135 , @xmath136 and @xmath137 ) are shown in fig . 1 , together with an ad hoc fit to be discussed below .
the two correlation functions @xmath138 and @xmath139 have to be understood as averages over two correlators involving mutually parallel components of @xmath140 or @xmath141 orthogonal to @xmath142 . for better visibility
the sum is shown in fig .
the data shown are averaged over correlators with exactly equal distance .
even then the data still would show an eventual rotational non - invariance ( in the three - dimensional distance ) if rotational symmetry would be seriously broken . at first , analogously to the procedure of digiacomo et al .
applied to their @xmath3 cooled data @xcite , we have adopted the decomposition ( [ eq : survey ] ) into the basic structure functions for the @xmath140 and @xmath141 correlators separately , here however without a perturbative part which is suppressed by the steps of blocking and inverse blocking ( see sec .
we have assumed the following form for the basic @xmath140 and @xmath141 correlators [ eq : pure_exp ] d(x ) & = & a_0 e^-|x|/_0 , + [ -.2 cm ] + [ -.25 cm ] d_1(x ) & = & a_1 e^-|x|/_1 .
only correlator data at distances @xmath143 have been included into the fit , in accordance to our understanding that fields below that scale are interpolated in a smooth way and taking the extendedness of the field strength operator ( [ eq : gemunu ] ) into account .
the four corresponding parameters are given in table ii to allow a comparison with the amplitudes and exponential correlation lengths obtained for @xmath144 by the cooling method for @xmath2 @xcite as analyzed in ref .
@xcite .
first , we mention that all structure functions are found positive at these temperatures in both cases . this property is corroborated also in the second way of fitting to be discussed below .
generally , at all temperatures , the two - exponential fit does not properly describe the decay of @xmath145 as measured here by the smoothing method at the largest distances .
due to cancellation in the expression @xmath146 this function would drop too fast .
the two correlation lengths associated with @xmath75 and @xmath147 by the fit are almost degenerate and do not grow with temperature . for the @xmath140 correlators ,
a correlated fit of @xmath148 and @xmath149 starts to be increasingly inadequate at temperatures @xmath150 .
an indication is the @xmath140 correlator amplitude @xmath151 for @xmath152 where the signal has an uncertainty of the same order of magnitude .
taking the parameters of this fit seriously , the most dramatic feature is the drop of the correlation length of @xmath73 across the phase transition . at @xmath132 ( @xmath153 ) , safely below the deconfining transition , we observe a weak splitting of the two correlation lengths , with @xmath154 fm . at @xmath155
we find still @xmath156 fm . immediately above the deconfining transition ( at @xmath157 )
it has dropped already to @xmath158 fm .
for the purpose of the fit shown in fig . 1 , we assumed all four functions to have a form like @xmath159 and required the same limit for @xmath160 for @xmath77 and @xmath78 , as well as for @xmath79 and @xmath80 .
in addition to this , have we considered remaining perturbative contributions to these four functions being proportional to @xmath161 .
for all temperatures and all correlation functions we found this contribution to be smaller than the signal by one or two orders of magnitude over the range of distances from 1 to 6 lattice spacings .
of course , also here only correlator data at distances @xmath143 have been included into the fit . in the following we therefor only discuss parameters which were determined by fits neglecting such a power - like perturbative contribution .
the parameters @xmath162 , @xmath163 , and @xmath164 are listed in table iii in lattice units . in the confinement phase
, the coefficient @xmath164 is relatively important only for @xmath165 and @xmath166 . while it remains roughly unchanged also at higher temperature for the magnetic correlator , it rises dramatically in the electric case in order to account for the rapid decay .
on the basis of this fit , we have determined the structure functions @xmath73 , @xmath74 , @xmath75 and @xmath76 which are shown in fig . 2 for all @xmath32 values .
in a first step the @xmath62 have been numerically obtained by integrating the difference @xmath167 ( as represented by the fits ) from @xmath168 to @xmath169 .
the integration constants @xmath170 are fixed by the condition that @xmath171 for @xmath172 .
we find the functions @xmath62 positive .
finally , the @xmath61 have been obtained subtracting @xmath62 from @xmath173 .
the amplitudes @xmath174 , @xmath175 , @xmath176 , @xmath177 ( which are related to the gluon condensate at high temperature ) and the integrated correlation lengths of @xmath73 and @xmath75 , @xmath178 , are given in table iv for the six temperatures .
as long as @xmath179 the functions @xmath62 are smaller by roughly a factor four than the corresponding @xmath61 .
that means that selfdual or antiselfdual fields are less dominant in this range of temperatures compared to zero temperature .
the ratio @xmath180 has been found @xcite considerably smaller below the deconfinement transition in an analysis of @xmath2 correlators obtained with cooling .
since there are no cooling data for @xmath0 so far in our temperature range , we can not decide whether this is due to the other gauge group or whether there is a systematic effect of cooling to enhance the contribution of locally ( anti)selfdual configurations to the correlator at higher temperature but still in the confined phase .
the condensates @xmath174 and @xmath176 only slightly grow across the transition and start to rapidly grow not before @xmath181 . up to these temperatures
the @xmath182 have increased to a size comparable with @xmath170 .
the integrated correlation lengths of @xmath61 shrink in a rather continuous way over the temperature range considered from 0.3 fm ( electric and magnetic ) at @xmath183 to 0.13 fm ( magnetic ) and 0.07 fm ( electric ) at @xmath184 ( fig .
the smooth change of the four structure functions with increasing temperature is shown in fig .
at first , we find our data approximately @xmath17 rotational invariant also at @xmath185 as long we consider the confinement phase . the small difference between the corresponding electric and magnetic correlators ( both for @xmath186 and @xmath187 ) probably can be attributed to overlapping fields of ( anti)calorons , within a small density expansion .
this picture , however , breaks down in the deconfined phase . in the confinement phase ,
the correlator @xmath188 practically coincides with @xmath189 ( see fig . 1 and note there the upward shift by a factor two in the latter ) . at the transition ,
the breakdown of @xmath188 ( in the correlation length ) is the most notable result of our analysis . comparing the two methods of fitting we find that the condensates estimated from the constrained , purely exponential fits in terms of two structure functions @xmath73 and @xmath190 tend to be bigger than the condensates from the second fit .
this is because the latter better accounts for the real shape of the correlators near zero distance .
the exponential fit also does not properly describe the correlator @xmath145 at large distances .
this observation holds for all temperatures . within the deconfined phase
, the @xmath140 correlators can not be described at all , by an ansatz in terms of exponential @xmath73 and @xmath190 functions .
thus , the rapid change of the exponential correlation lengths ( in particular , of the electric correlation length ) at the transition merely reflects the difficulty to reach a constrained fit within these assumptions .
the gauss - exponential fit proves to be much more robust over the whole range of temperatures . without any further assumptions concerning the shape of @xmath61 and @xmath62 ,
such a fit gives access to these functions which indeed seem to be poorly described by exponentials .
we have to describe them by integrated correlation lengths which can not be directly compared with the exponential correlation lengths of the supposedly exponential fits . in general
, the integrated correlation lengths turn out to be bigger than the exponential ones . in contrast to the exponential correlation lengths of the constrained exponential fits , the integrated correlation lengths associated to the correlation functions @xmath61 and @xmath62 , as reconstructed from the second fit , change rather smoothly . as the result of these smooth changes with temperature ,
the correlation function @xmath188 drops ( in correlation length ) at the deconfinement transition .
the reconstruction of the absolute size of @xmath8 and @xmath9 structure functions ( condensates ) from the fits of @xmath186 and @xmath187 becomes increasingly difficult at higher and higher temperature . below the deconfinement temperature and even near to @xmath7 as a characteristic feature of the data ,
it is well - recognized , that the @xmath9 part of the electric and the magnetic correlator is considerably smaller than the @xmath8 part .
the dominance of ( anti)selfdual fields , however , associated with this observation is already weaker in our present results , at temperatures below the transition , than it has been found for zero temperature .
it is also weaker than what has been found analyzing cooling data for quenched @xmath2 near the deconfinement temperature . in table v this statement can be checked for the so - called non - abelianicity of the magnetic correlator .
it is roughly 80 % at @xmath183 and drops to than 70 % at @xmath7 . in the deconfined phase @xmath191
amounts to approximately 50 % , with an increasing uncertainty .
the positivity of magnetic and electric @xmath9 is a clear feature of the data .
this applies to the correlators obtained in this work by smoothing as well as to the measurements using the cooling method ( for quenched @xmath2 ) . within a caloron analysis ( which successfully works only below @xmath7 for the @xmath2 cooling data )
there is no compelling reason to have @xmath47 .
this is unlike the instanton liquid description for @xmath3 .
a caloron analysis of the present data will be given elsewhere @xcite .
the sum of @xmath192 and @xmath193 is given by the extrapolation to zero distance of both @xmath186 and @xmath187 for the electric and magnetic correlators .
this quantity is known without big uncertainty and is also presented in table v in physical units for the six temperatures . the behavior of condensates and correlation lengths is sketched graphically across and beyond the deconfinement transition in fig .
quantitatively , the correlation lengths are not very different from the results of the cooling method , applied on @xmath2 configurations . the present work dealt with @xmath0 pure gauge theory instead of the phenomenologically more interesting case of @xmath2 .
however it has shown an interesting pattern of temperature dependence of the magnetic and electric correlation lengths related to the different structure functions .
while the basic structure functions @xmath8 and @xmath9 change smoothly across the phase transition , the most dramatic effect , the breakdown of @xmath165 , results from the interplay . a simple picture according to which @xmath73 would turn to zero in the deconfined phase is not corroborated by our data
. it seems to be worthwhile to investigate the field strength correlators for @xmath194 using the cooling method , too , in order to compare it with the smoothing method applied here
. it would be even more interesting to implement the rg smoothing method for the gauge group @xmath2 .
this would then allow for a direct comparison with the pioneering results already available thanks to the pisa group .
p. v. landshoff and o. nachtmann , z. phys .
c * 35 * , 405 ( 1987 ) ; o. nachtmann , ann .
( ny ) * 209 * , 436 ( 1991 ) ; a. krmer and h. g. dosch , phys .
* b252 * , 669 ( 1990 ) ; h. g. dosch , e. ferreira , and a. krmer , phys .
d * 50 * , 1992 ( 1994 ) ; a. f. martini , m. j. menon , and d. s. thober , phys .
d * 57 * , 3026 ( 1998 ) . | we summarize recent attempts to extract characteristics of non - perturbative vacuum structure from lattice measurements of the gauge invariant field strength correlator . as an alternative to cooling , we propose to apply the renormalization group ( rg ) smoothing method in lattice studies . for pure @xmath0
gauge theory we present magnetic and electric correlation lengths and condensates related to various correlators over a temperature range of @xmath1 .
+ pacs : 11.15.ha , 12.38.gc , 11.15.kc , 11.10.wx |
40004000 = 1000 = ifundefinedreset@font@font footnotetext#1 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbcitea citea , m tempcnta = tempcntbtempcnta # 1 o p # 1#1 # 1 # 1 # 1 # 1 # 1#2 # 1#2_#1f_#2 # 1#1 # 1#1 _ # 1@xmath1 # 1 _ 0
recently , the quantum properties of @xmath0 supersymmetric field theories have been under active investigation .
in particular , seiberg and witten @xcite analyzed an asymptotically free @xmath2 yang - mills theory and determined the leading contribution of the low energy effective action for all values of the gauge coupling .
it turns out that in the strong coupling region no massless @xmath2 gauge bosons exist ; instead magnetic monopoles become massless and the effective low energy theory is best described by a weakly coupled dual @xmath3 magnetic gauge theory .
these developments lead to intensive attempts to uncover the same physical phenomenon in string theory and study its gravitational and `` stringy '' generalizations .
it has been conjectured that for strongly coupled heterotic @xmath0 vacua the `` dual '' description is provided by weakly coupled @xmath0 vacua of type ii strings .
this conjecture is supported by a number of concrete `` pairs '' of dual vacua @xcite where the perturbative effective action of the heterotic vacuum ( at least , for the vector multiplets ) has been matched with the dual type ii vacuum in an appropriate expansion @xcite .
furthermore seiberg witten theory should appear in this context @xcite and can be recovered in the @xmath4 limit @xcite .
these encouraging results make it interesting to go beyond a model by model analysis and study more generally the properties of the conjectured duality .
however , for an arbitrary heterotic string vacuum it is presently unknown how to construct the dual partner .
one might suspect that this duality in 4 dimensions is a consequence of string - string duality @xcite in six dimensions and the dual pairs of @xcite are indeed constructed in this way .
it is not clear how to show the connection for the pairs of @xcite however .
initial steps in this latter case were taken in @xcite where it was noticed that both the type ii string and the heterotic string could be `` fibred '' over a @xmath5 .
that is , the manifold , @xmath6 , on which the type iia string was compactified could be written as a fibre bundle with base @xmath5 and generic fibre a k3 surface where the heterotic string was compactified on k3@xmath7 which can be written as a bundle with base @xmath5 and generic fibre @xmath8 .
it is then tempting to try to use string - string duality fibre - wise to map the type iia string on the k3 surface into a heterotic string on @xmath8 .
the relevance of k3-fibrations was first noticed in @xcite where it was shown that such a structure fits nicely with properties of the dual heterotic string compactified on k3@xmath9 .
( various related aspects of k3-fibrations were also discussed in @xcite . ) in this paper we will take this point further and argue that the appearance of the k3-fibration is actually unavoidable .
the argument will borrow results from the `` phase '' picture of the moduli space of @xmath0 conformal field theories @xcite .
in particular , we need to understand clearly what we mean by an object that is recognizably a heterotic string and an object that is recognizably a type iia string theory compactified on a manifold . in the latter case , we know from the phase picture that we need to identify a `` '' phase of the moduli space and we will see that we must also identify a `` weakly - coupled '' phase in the heterotic string moduli space .
the phase picture allowed a complete resolution of the question of manifolds without mirror partners @xcite .
we will see that exactly the same considerations apply to heterotic strings without dual type ii partners and _ visa versa_. if we demand that we do have a dual pair in which the weakly - coupled phase of a heterotic string maps into the phase of a type iia string then we will see that this manifold must have a k3-fibration .
one bound which appears naturally on the heterotic side is on the rank of the gauge group .
one can use conformal field theory considerations to show that it can not exceed 24 .
it has been troubling that no such bound is known on the type ii side .
we will see how k3-fibrations give a natural interpretation to this effect although it appears that this bound can be broken by nonperturbative effects on the heterotic side .
c let us start by summarizing the generic properties of perturbative heterotic @xmath0 vacua .
the heterotic string is based on a conformal field theory with central charge @xmath10 . out of this total central charge a @xmath11 `` block '' is used to build the four - dimensional space - time while the left over @xmath12 conformal field theory is only further constrained by the amount of space - time supersymmetry and the size of the gauge group .
@xmath0 space - time supersymmetry requires that the right moving internal @xmath13 conformal field theory splits into a free complex boson ( of central charge @xmath14 ) and a @xmath15 conformal field theory with @xmath16 world - sheet supersymmetry @xcite .
the gauge group @xmath17 arises from the left moving @xmath18 conformal field theory and the free @xmath19 boson .
the latter gives rise to two abelian @xmath3 gauge bosons which are identified with the graviphoton and the vector partner of the dilaton .
the @xmath18 conformal field theory generates an arbitrary ( possibly non - abelian ) gauge group @xmath20 of maximal rank 22 .
therefore @xmath21 ^ 2 $ ] and the rank of @xmath17 is bounded by @xmath22 the supersymmetric partners of the gauge bosons are two gauginos and a complex scalar all in the adjoint representation of @xmath17 ; they form what is called an @xmath0 vector multiplet .
the matter fields usually transform in the fundamental representation of @xmath17 and they reside in @xmath0 hypermultiplets which contain two weyl fermions and four real scalars .
all string vacua also feature gauge neutral scalars termed moduli which are exactly flat directions of the effective potential and which parameterize the perturbative degeneracy of a given vacuum family . such moduli arise as the scalars of either vector or hypermultiplets and they can be thought of as parameters taking values in a moduli space .
@xmath0 supersymmetry implies that the moduli space @xmath23 is always locally a product @xcite @xmath24 where @xmath25 ( @xmath26 ) is spanned by the scalars in the vector multiplets ( hypermultiplets ) .
the scalar fields @xmath27 in the cartan subalgebra of @xmath20 are flat directions of the effective potential . if all @xmath28 s are turned on ( a generic point in the moduli space
) @xmath20 is broken to its maximal abelian subgroup ( which is at most @xmath29^{22}$ ] ) while at special points in the moduli space this gauge symmetry can be enhanced to a non - abelian group .
in addition to the @xmath28 moduli there is also always the string dilaton present in the massless spectrum .
its vacuum expectation value determines the string coupling constant and organizes the string perturbation theory .
together with the dual axion of an antisymmetric tensor field it forms a complex scalar @xmath30 of an abelian @xmath3 vector multiplet gauge bosons arises in the @xmath31 block of the conformal field theory . ] and in the standard heterotic normalization one defines @xmath32 such that large @xmath30 corresponds to weak coupling . with this convention the peccei
quinn symmetry associated with the axion is given by a shift @xmath33 where @xmath34 is a continuous parameter in string perturbation theory .
however , once nonperturbative corrections are taken into account this continuous symmetry is broken down to a discrete subgroup and then @xmath34 must be an integer .
the couplings of the vector multiplets ( the khler potential and the gauge couplings ) are entirely encoded in a holomorphic prepotential @xmath35 . due to the non - renormalization theorem of @xmath0 supersymmetry @xmath35
does not receive quantum corrections beyond one - loop in perturbation theory but can ( and is ) corrected nonperturbatively . using the fact that the dilaton counts string loops , @xmath35 can be expanded for large @xmath30 ( weak coupling ) according to @xmath36 where @xmath37 @xmath38 is the tree level ( one - loop ) contribution while @xmath39 denotes nonperturbative corrections .
the universal couplings of the dilaton and the continuous peccei quinn symmetry of the axion completely determine @xmath37 for all heterotic string vacua to be @xcite ( using the conventions of @xcite ) @xmath40 where @xmath30 and @xmath28 are the `` special '' @xmath0 coordinates .
@xmath41 is necessarily @xmath30-independent .
the tree level prepotential given in ( [ eq : fhettree ] ) corresponds to the khler potential @xmath42 which is the khler potential of the homogeneous space @xmath43 up to discrete identifications . note that ( [ eq : mhet ] ) is subject to string loop corrections and , in general , is not isomorphic to @xmath25 .
so far we concentrated on the lowest order ( two derivative ) couplings in the effective action .
there is a special class of higher derivative curvature couplings which arise from chiral integrals in @xmath0 superspace and therefore are also governed by holomorphic functions @xmath44 of the vector multiplets .
these are couplings of the form @xmath45 where @xmath46 is the ( anti - self - dual ) riemann tensor , @xmath47 is the field strength of the graviphoton and the couplings @xmath48 obey @xcite always is a holomorphic quantity while @xmath48 is the non - harmonic coupling . ]
@xmath49 at the string tree level @xmath50 holds and therefore @xmath51 is a harmonic function of the vector moduli . however , at one - loop @xmath51 develops a holomorphic anomaly @xmath52 . for @xmath53
one finds @xcite @xmath54 where @xmath55 is the khler potential defined in ( [ eq : kahler ] ) ; @xmath56 ( @xmath57 ) counts the number of hypermultiplets ( vector multiplets ) .
similar to ( [ eq : fhet ] ) the holomorphic @xmath58 s can be expanded in the dilaton according to @xmath59 again @xmath60 is dilaton independent and the perturbative peccei
quinn symmetry of the axion determines @xmath61 where the normalization of @xmath62 is fixed by the normalization of ( [ eq : anomaly ] ) .
type ii string vacua are built from conformal field theories of central charge @xmath63 .
the four - dimensional space - time degrees of freedom arise from a @xmath64 block leaving @xmath65 as the internal conformal field theory .
the standard way to obtain @xmath0 space - time supersymmetry is to demand a left - right symmetric @xmath0 world - sheet supersymmetry of the internal @xmath65 conformal field theory .
an example of such a conformal field theory is provided by the on a calabi - yau manifold .
let us consider a type iia superstring compactified on a manifold @xmath6 .
equivalently , we can compactify the type iib string on @xmath66 , the mirror of @xmath6 . the number of vector- and hypermoduli is directly related to the hodge numbers of the calabi yau compactification .
specifically one finds @xmath67 for type iia and @xmath68 for type iib ; in both cases the additional hypermultiplet corresponds to the type ii dilaton .
the gauge group @xmath69 always is a product of @xmath70 abelian @xmath3 factors and therefore one has in general @xmath71 where the `` @xmath72 '' counts the graviphoton .
the dilaton always is the member of a hypermultiplet and thus ( [ eq : modulispace ] ) implies that for type ii vacua @xmath25 is independent on the dilaton and therefore determined at the string tree level exactly with no further corrections perturbatively or nonperturbatively .
near the large radius limit of @xmath6 , the position in the moduli space @xmath25 is given by the complexified khler form on @xmath6 that is , an object of the form @xmath73 , where we expand @xmath74 so that @xmath75 and @xmath76 are real numbers and @xmath77 represents a basis of @xmath78 ( we assume that @xmath79 ) . as @xmath80 is dual to @xmath81 and @xmath81 is dual to @xmath82 , we have an isomorphism between @xmath80 and @xmath82 . using this isomorphism , we can associate a divisor , @xmath83 , of @xmath6 to each @xmath77 . in the large radius limit ( or large complex structure limit for type iib )
the holomorphic prepotential has the generic structure @xmath84 where @xmath85 denotes the moduli of the vector multiplets in @xmath0 special coordinates and we have omitted the loop terms .
recall @xcite that the instanton corrections in ( [ eq : ftt ] ) come from rational curves within @xmath6 .
as these curves get larger , the instanton effects become smaller . in order for ( [ eq : ftt ] ) to make sense , we must assume that the sum produced by the instantons is convergent .
if it is not , one can often replace the model by another theory ( such as a landau - ginzburg theory ) in which some other instanton sum converges .
this is regarded as the underlying @xmath0 superconformal field theory being in another `` phase '' @xcite .
if we insist that we are in a phase and thus that ( [ eq : ftt ] ) is a convergent series then we are demanding that all the rational curves on @xmath6 are sufficiently large .
now we are interested in identifying those calabi
yau compactification which can serve as possible dual descriptions of heterotic vacua .
the idea is , that once all nonperturbative effects have been taken into account , the heterotic theory and the type ii theory describe exactly the same physics in the infrared .
one test of the proposed duality is to check that both theories have identical moduli spaces .
however , in either case we only know the respective weak coupling limits . nevertheless due to the product structure of the moduli space and the fact that the dilaton resides in different super - multiplets in the two theories it is possible to compare the heterotic @xmath25 in a weak coupling expansion with the exact @xmath25 of the type ii vacuum .
in particular , one of the vector moduli @xmath86 in the type iia vacuum must be the image of the axion - dilaton pair of the heterotic string .
let us denote this particular member by @xmath87 and let @xmath88 be the associated divisor .
is a `` @xmath89-cartier '' divisor .
this follows since an irrational class would not be compatible with the peccei
quinn symmetry . ]
as we have just said we can only compare the moduli spaces for large @xmath30 and hence @xmath90 also has to be large . in this limit
the type iia @xmath91-field is a periodic variable and obeys @xmath92 .
this is consistent with the peccei
quinn symmetry @xmath93 of the heterotic axion if we identify @xmath94 using the fact that both the complexified khler form and the axion - dilaton are special @xmath0 coordinates ( [ eq : ax - ii ] ) is exact up to symplectic @xmath95 reparametrizations and not only valid at the large radius limit .
the other vector multiplets will be denoted by @xmath96 ( or equivalently @xmath97 ) , for @xmath98 .
given our assumption that the weakly coupled phase of the heterotic string maps into the phase of the type iia string we can obtain from ( [ eq : fhettree ] ) and ( [ eq : ftt ] ) that @xmath99 @xmath100 must be an element of the second homology of @xmath6 . since ( [ eq : ssi ] ) tells us that this element has zero intersection with any 4-cycle , it must be trivial .
that is , @xmath101 we know that @xmath88 itself is not trivial as the associated modulus is not trivial . in the language of algebraic geometry we have shown that the `` numerical @xmath102-dimension '' of the divisor @xmath88 is one .
next , let us analyze the positivity properties of @xmath88 .
we have asserted that we are in the phase of the type iia string . from the work of @xcite
it follows that we know something about @xmath103 .
very roughly , it means that @xmath103 lies within the khler cone of @xmath6 .
that is , the volume of any algebraic subspace within @xmath6 is measured by the khler form to be positive .
suppose we take a theory which is in the phase and modify the khler form as @xmath104 where @xmath105 is a large real number .
in the heterotic language , this amounts to increasing the value of the dilaton , i.e. , making the string more weakly coupled .
thus , the instanton sums will become more convergent . if this is to have the same effect in the type iia string , then this change in the khler form must keep us inside the khler cone of @xmath6 .
this amounts to the assertion that @xmath106 for any algebraic curve , @xmath107 , within @xmath6 . in the language of algebraic geometry , this tells us that @xmath88 is `` nef '' .
lastly we need one more property of @xmath88 which follows from identifying the higher derivative curvature couplings @xmath58 in both theories .
since the @xmath58 s depend only on the vector multiplets they are independent on the type ii dilaton and one finds that they arise solely at the @xmath108-loop order . in the large radius limit one obtains @xcite @xmath109 where @xmath110 denotes the second chern class of @xmath6 .
furthermore , @xmath111 develops a holomorphic anomaly given by @xcite @xmath112 where @xmath113 is the ricci - tensor on the moduli space .
let us first note that for @xmath114 the large @xmath30 limits of ( [ eq : fnhettree ] ) and ( [ eq : fgtt ] ) already agree with no further conditions imposed .
however , identifying the @xmath115 s of the dual pairs results in a constraint on @xmath116 .
note that the holomorphic anomaly of ( [ eq : anomalyii ] ) simplifies if , in the large @xmath103 limit , the type ii prepotential and khler potential are identical with the heterotic @xmath35 and @xmath55 of ( [ eq : fhettree ] ) and ( [ eq : kahler ] ) .
a straightforward computation shows that ( [ eq : anomalyii ] ) reduces in this limit to @xmath117 where @xmath118 is defined in ( [ eq : anomaly ] ) .
we see that the anomalies of a dual pair coincide up to the overall normalization . in turn
this can be used to align the overall normalization between the @xmath115 s in each vacuum . identifying the @xmath115 s using ( [ eq : ax - ii ] ) , ( [ eq : fnhettree ] ) , ( [ eq : anomaly ] ) , ( [ eq : anomalyiib ] ) and ( [ eq : fgtt ] ) results in @xmath119 we may now use the results of oguiso @xcite which state
that let @xmath6 be a minimal threefold .
let @xmath102 be a nef divisor on @xmath6 .
if the numerical @xmath102-dimension of @xmath102 equals one and @xmath120 then there is a fibration @xmath121 , where @xmath122 is @xmath5 and the generic fibre is a k3 surface . we have obtained the desired result @xmath6 has to be a k3-fibration in order to be the dual partner of a heterotic vacuum .
note that in general the fibre is allowed to degenerate to something other than a k3 surface over a finite number of points on @xmath122 .
this fibration is shown in figure [ fig : fib ] .
we should emphasize two points about our derivation of the result .
firstly , we assumed that the manifold in the type iia theory was in the phase when the heterotic string was weakly - coupled .
it is certainly conceivable that there may be dual pairs where the phases are not aligned like this . in such an example , there is then no reason why a manifold on the type iia side , which appears for some strongly - coupled regime of the heterotic moduli space , should be a k3-fibration . in other words ,
the weakly - coupled heterotic string might be dual to a type iia string theory on a manifold @xmath6 , which is not a k3-fibration , in the following sense .
the instanton sum in ( [ eq : ftt ] ) is necessarily divergent but could be reinterpreted in terms of some other non- phase ( such as a perturbed landau - ginzburg theory or orbifold @xcite ) to give a convergent sum of corrections .
secondly , we are stating only that the type iia string is compactified on a k3-fibration .
the type iib string is compactified on @xmath66 , the mirror of @xmath6 , which may or may not be a k3-fibration .
the question as to what extent mirror symmetry acts within the class of k3-fibrations has not been analyzed at this point in time so we can not make any strong statements about when @xmath66 is a k3-fibration .
one can observe though that in the cases discussed in @xcite , where the mirror map is in the greene - plesser orbifold form @xcite , then the mirror orbifolding operation preserves the fibration structure and so both @xmath6 and @xmath66 are k3-fibrations . note that ( [ eq : ctwo ] ) not only tells us that @xmath123 but that the value is the euler characteristic of a k3 surface .
this tells us that the element in @xmath82 corresponding to @xmath88 is given precisely by a generic k3 fibre @xcite .
clearly the 2-cycle dual to this is given by @xmath122 , the base space .
it then follows that @xmath124 is the cohomology element that gives us the _ size _ of this 2-cycle .
that is , the dilaton in the heterotic string is given by the size of the base @xmath5 in the fibration of @xmath6 .
thus far we have made no appeal to string - string duality in six - dimensions but we can use this duality to reproduce the result we have just found concerning the dilaton . let us set the axion in the heterotic string to zero for simplicity . in @xcite
it was shown that @xmath125 where @xmath126 is the dilaton in six dimensions for the heterotic string and @xmath127 is its metric with the corresponding values primed for the type iia string .
let us apply this to our picture in four dimensions by `` compactifying '' the six - dimensional space - time over @xmath5 .
given the weyl - rescaling between the type iia and heterotic space - times , the areas of these rational curves will differ by a factor of @xmath126 .
the action for the heterotic string can then be written @xmath128 this last form of the action tells us that the area of the base in the type iia picture gives the four - dimensional dilaton of the heterotic string as we saw above .
for the dilaton modulus in the type iib picture we can use the monomial - divisor mirror of @xcite ( or , more specifically , the results of @xcite ) to find the corresponding deformation of complex structure of @xmath66 .
note also that we know the result exactly since the dilaton is given by the khler form , the map between the dilaton and the complex modulus of @xmath66 is corrected by solutions to the picard - fuchs equation .
for example , in @xcite @xmath66 was taken to be an orbifold of the hypersurface @xmath129 in @xmath130 .
it was then conjectured that there was a heterotic string dual to a type iib string compactified on @xmath66 . from what we have said above and using the results of @xcite
it follows that the heterotic dilaton is given by @xmath131 where @xmath132 this agrees with the conjectured relationship in @xcite .
one can show that the example with @xmath133 of @xcite is also consistent with our statements .
it is perhaps worth emphasizing that we have deduced ( [ eq : eg1 ] ) given only the existence of the weakly - coupled heterotic string no further knowledge of the heterotic string ( such as enhanced gauge symmetries ) was required .
let us now turn to the other @xmath134 vector moduli .
now that we know that @xmath6 is a k3-fibration , we can say exactly where the other contributions to @xmath135 come from .
the general statement is that they can be divided into two classes those from the generic fibre and those from the singular fibre .
let us concentrate first on the contribution from the generic fibre .
let @xmath136 be an algebraic curve in a generic k3 fibre .
as we move about the base space we can map the curve into equivalent curves in the other generic fibres .
adding all these curves together gives us a divisor , @xmath97 in @xmath6 .
that is , we have written @xmath97 as a fibration over @xmath122 where @xmath136 is the generic fibre .
( note that @xmath97 is not dual to @xmath136 . )
thus , a curve in the generic fibre contributes to @xmath137 . because a generic k3-fibration has monodromy , we may have two curves in a generic fibre that can be mapped into each other by cycling round a path on @xmath122 .
clearly such a pair only contribute one to @xmath137 as they both were required in building @xmath97 .
let us denote a generic fibre @xmath138 , where @xmath139 .
the set of ( duals of ) algebraic curves in @xmath138 generate a sublattice of @xmath140 called the `` picard lattice '' of @xmath138 .
clearly both @xmath141 and the lattice @xmath140 can be embedded in @xmath142 .
the picard lattice can be considered the intersection @xmath143 .
consider the space @xmath144 .
the natural wedge product between elements in this space endows it with a metric .
one can show that this metric has signature ( 3,19 ) ( see , for example , @xcite ) .
let @xmath145 be a @xmath146-form on @xmath138 .
one can show that the real and imaginary parts of @xmath145 span a space - like 2-plane in @xmath147 . as any element of the picard lattice is
a ( 1,1)-form and so lies orthogonal to this 2-plane , the picard lattice must be embeddable in @xmath148 .
the `` picard number '' of a k3 surface is the rank of the picard lattice .
clearly then , the picard number can be no bigger than 20 .
the khler form on a k3 surface is a vector within the real space generated by the picard lattice .
the volume of the k3 surface is given by the length of this vector which must therefore be positive .
thus we have shown that the picard lattice must have signature @xmath149 . from what we have said above ,
the vector multiplets coming from the generic fibre are the generators of the monodromy - invariant subspace of the picard lattice .
this subspace must contain the khler form of the fibre and so also has signature @xmath149 .
the rank is clearly less than or equal to the picard number .
consider the intersection numbers @xmath150 where the latter two divisors come from curves in the generic fibres .
this is equal to the intersection @xmath151 within the fibre and is thus given by the intersection form of the monodromy - invariant picard lattice .
this means that we have exactly reproduced the heterotic string result ( [ eq : fhettree ] ) in terms of the type iia string .
we also know that there is at least one ( the volume of the fibre ) but no more than 20 moduli from this source .
adding these to the dilaton and the @xmath3 from the gravity multiplet shows that the rank of the gauge group for k3-fibrations with only contributions to @xmath137 from the base space and generic fibers satisfies @xmath152 which is consistent with the heterotic bound ( [ eq : hetrank ] ) .
the upper limits are suspiciously close to each other and we can probably bring these results into agreement once we take into account the _ quantum _ geometry of the k3 fibre . in the above analysis we used purely classical geometry for the analysis of the picard lattice . given the appearance of quantum effects in k3 surfaces @xcite it is not at all unreasonable to expect that some quantum picard lattice exists for a planck - sized k3 surface with picard number 22 .
this would bring the upper limit of @xmath69 into complete agreement with the heterotic limit
. the lower limit of the heterotic vacuum can never be reached for a calabi
yau manifold since there necessarily has to be a volume form @xmath103 which obeys @xmath153 and thus can not be the image of the heterotic dilaton .
however it is conceivable that a model which does not have a phase may be used to achieve this lower limit .
so far we have discussed the successes of the type iia / heterotic duality picture .
now let us discuss some of the short - comings and the reasons why they must appear .
the first question which comes to mind concerns the contributions to @xmath135 from the degenerate fibres .
let us call such a divisor @xmath154 . clearly such a divisor will not intersect the generic fibre @xmath88 and so @xmath155 in contradiction with ( [ eq : fhettree ] ) for the heterotic string .
this implies that @xmath156-fibrations where @xmath135 contains moduli corresponding to degenerate fibers can not be the dual of a standard weakly coupled heterotic string vacuum as we described it in section 2 .
however , such fibrations still have a candidate for the image of the heterotic dilaton but there seem to be no heterotic states which @xmath154 can be the image of .
if there are such states ( [ eq : decoup ] ) implies that they decouple at the tree level and only arise at one - loop and/or nonperturbatively ( equivalently this says that they can not be states in the conformal field theory ) . since in general @xmath157 the corresponding heterotic states would have to arise already at the one - loop level .
alternatively they could be states that become massless only nonperturbatively but with `` unusual '' couplings to the dilaton such as the ramond - ramond states in type ii strings .
being members of vector multiplets they would enhance the rank of @xmath17 nonperturbatively but so far no such states have been identified directly in the heterotic string
. the number of states coming from singular fibres can be very large easily taking the rank of the gauge group beyond the expected limit of 24 .
but as we have just argued these vector multiplets have no conformal field theory interpretations on the heterotic side and thus no contradiction arises .
conversely , demanding the absence of states corresponding to degenerate fibers implies the bound ( [ eq : iirank ] ) perfectly consistent with the heterotic string .
what if @xmath6 is not a k3-fibration ? what happens to the heterotic dual ?
this question is remarkably similar to the question of if a manifold has a mirror partner .
a simple example of a manifold without a mirror partner is that of one with no deformations .
this implies @xmath158 and so the mirror would have @xmath159 which is impossible for a khler manifold .
it turns out however that even in many cases where @xmath160 , the manifold may have no mirror .
this was analyzed in @xcite where it was shown that some @xmath0 superconformal field theories , appearing as the mirror of a theory , have a non - trivial moduli space of ( 1,1)-forms but nowhere in this moduli space is there a phase .
we claim that the same thing happens for type ii / heterotic duality .
the moduli space of vector multiplets in a type iia string may be non - trivial but it may be that there is simply no place in this moduli space where there is a weakly - coupled heterotic string .
this should happen if @xmath6 is not a k3-fibration .
of course , this works both ways .
given the moduli space of vector multiplets of a heterotic string there may be no place in this space where a weakly - coupled type iia string compactified on a manifold can be described .
it is worth emphasizing that missing dual partners can enter into the picture of connecting up the moduli space of @xmath0 theories via phase transitions as in @xcite and flops as in @xcite .
it may be that there is an extremal transition ( such as a conifold transition ) or flop on the type iia side from a manifold which is a k3-fibration to a manifold which is not . in this case
, the extremal transition or flop involves shrinking down the base @xmath5 of the fibration .
this means that on the heterotic side the phase transition occurs at strong coupling and the `` new '' phase no longer has a ( heterotic ) weak coupling limit .
we thank m. gross for many important contributions .
it is also a pleasure to thank p. argyres , p. berglund , s. chaudhuri , b. de wit , b. greene , j. harvey , v. kaplunovsky , d. morrison , j. polchinski , r. schimmrigk and a. strominger for useful conversations as well as all the other `` plancksters '' at the i.t.p .
, santa barbara program where part of this work was done under the nsf grant phy94 - 07194 .
we are especially grateful to p. langacker , p. nath and j. polchinski for organizing a stimulating workshop .
the work of p.s.a . is supported by a grant from the national science foundation .
the work of j.l . is supported by a heisenberg fellowship of the dfg and the german - israeli foundation . | we consider the general case of @xmath0 dual pairs of type iia / heterotic string theories in four dimensions .
we show that if the type iia string in this pair can be viewed as having been compactified on a calabi - yau manifold in the usual way then this manifold must be of the form of a k3 fibration .
we also see how the bound on the rank of the gauge group of the perturbative heterotic string has a natural interpretation on the type iia side .
epsf # 1date_#1 _ # 1thefnmarkfootnotetext1991 _ mathematics subject classification .
_ # 1 # 1thefnmarkfootnotetext _ key words and phrases .
_ # 1 ps .
@firstpageps .
@empty oddhead evenheadoddhead # 1 |
the continuing search for new superconductors has recently resulted in the identification of a new family of superconducting phosphide compounds with the chemical composition _ _ a__pt@xmath0p ( _ a _ = sr , ca , or la).@xcite these materials , whose superconductivity has been claimed to be driven by conventional electron - phonon interactions , adopt a distorted anti - perovskite structure , resembling the structure of several noncentrosymmetric superconductors , such as lapt@xmath0si or cept@xmath0si.@xcite contrary to the latter , however , srpt@xmath0p exhibits an inversion center due to the staggered arrangement of pt octahedra .
noncentrosymmetric superconductors , characterized by antisymmetric spin - orbit couplings , are currently the subject of intense research.@xcite therefore , a comparative study of their centrosymmetric counterparts is of particular interest .
indeed , the sr - based compound of the new family , with a critical temperature @xmath7k , has already been investigated in some detail .
the first study included measurements of the specific heat @xmath8 and hall resistivity @xmath9.@xcite the specific - heat data were interpreted as indicating a strong electron - phonon coupling with a relatively large ratio @xmath10 , and a fully - gapped excitation spectrum below @xmath4 with a zero - temperature gap value @xmath11 mev .
the nonlinear magnetic field dependence of hall resistivity was attributed to the presence of multiple fermi - surface pockets .
less details are known on lapt@xmath0p , whose critical temperature @xmath12k@xcite is distinctly lower than that of the sr compound , the difference being most likely due to the different valencies of the cations . there
exist several theoretical interpretations of the srpt@xmath0p data , which differ in their conclusions . in ref .
, the proximity to a dynamical charge - density wave instability is claimed to favor the occurrence of superconductivity in srpt@xmath0p . in another approach , based on a migdal - eliashberg - type analysis
, it is argued that a conventional phonon - mediated superconductivity is observed.@xcite density - functional - theory calculations@xcite suggest that the onset of superconductivity is due to a strong coupling between the @xmath13-hybridized bands with low - energy phonon modes confined in the @xmath14 plane of the crystal lattice .
the 2d character of these modes seems essential to preserve the antipolar arrangement of the distorted pt octahedra in srpt@xmath0p , thus enhancing both the electron - phonon coupling constant @xmath15 and , consequently , the critical temperature @xmath4 .
the only experimental investigation at a microscopic level cited in the literature focused on muon - spin rotation ( @xmath16sr ) measurements.@xcite the established temperature dependence of the penetration depth in the form of @xmath17 again indicated a single gap with a value of @xmath18mev .
however , an upward curvature of the upper critical field @xmath19 just below @xmath4 was interpreted as reflecting two - band superconducivity with equal gaps but differing values of the coherence lengths . in this work
we report on the microscopic physical properties of srpt@xmath0p , investigated by means of nuclear magnetic resonance.@xcite the measurements described below are based on @xmath20p nmr , and probe both static ( line widths and -shifts ) , as well as dynamic ( spin - lattice relaxation ) properties of the material .
overall the data indicate that srpt@xmath0p is a rather simple metal and that the pairing configuration is most likely of spherically - symmetric ( @xmath5-wave ) character . a comparison with the noncentrosymmetric superconductors lapt@xmath0si and cept@xmath0si is intended to clarify the peculiarities of the latter . the lack of inversion symmetry of the crystal lattice may partly account for the much lower @xmath4 of lapt@xmath0si , while the rare - earth cations seem responsible for the unusual electronic properties of cept@xmath0si.@xcite
polycrystalline samples of srpt@xmath0p were synthesized under conditions of high pressure and high temperature .
high - purity ( 99.99% ) coarse powders of sr , pt , and p were mixed in the stoichiometric 1:3:1 ratio , then thoroughly ground , and finally enclosed in a boron - nitride container . to avoid a possible degradation of the material due to air exposure
, all the preparatory steps were done in a glovebox under argon atmosphere . for the heat treatment ,
the bn crucible was placed into a pyrophyllite cube containing a graphite heater .
an external pressure of 2 gpa was applied at room temperature and kept constant during the two - hour ramping up of the temperature to 1050@xmath21c .
the final temperature was kept constant for 40 hours and then lowered back to room temperature within approximately one hour . after releasing the pressure ,
the polycrystalline samples were extracted and kept in closed capsules filled with ar gas during the corresponding experiments .
subsequently - recorded x - ray powder diffraction patterns agreed with those reported in ref . .
measurements of the low - temperature magnetization revealed the onset of a diamagnetic response at @xmath22k , again in good agreement with the originally reported value.@xcite the relevant magnetic susceptibility data @xmath23 are shown in fig . [
fig : magnetization ] where , at low temperatures , we observe a significant diamagnetic response , typical for the onset of superconductivity .
temperature dependence of the zero field - cooled ( zfc ) and field - cooled ( fc ) dc magnetization measured at @xmath24 mt . the sample shows a sizeable diamagnetic response with @xmath22k .
inset : structure of srpt@xmath0p showing the pt@xmath25 octahedra ( adapted from ref . ) .
the asymmetric positions of the apical pt(ii ) ions give rise to alternating electric polarization vectors ( blue arrows ) , hence preserving the centrosymmetric nature of srpt@xmath0p . ]
although all the constituent elements of srpt@xmath0p feature nmr - active nuclei , the @xmath1p nucleus is clearly the best choice with respect to isotopic abundance ( 100% ) and gyromagnetic ratio ( @xmath26 mhz / t ) .
in addition , the @xmath27 nuclear spin of @xmath1p helps avoiding complications due to quadrupole effects . in order to probe an adequate region of the superconducting part of the @xmath28-@xmath29 phase diagram above 1.6 k ( i.e. , accessible with a @xmath30he flow cryostat ) ,
the external field for the nmr studies was set to @xmath31 t , corresponding to a critical temperature of approximately 5k ( see , e.g. , fig . 1 in ref . ) .
analogous measurements probing the normal state in the same temperature regime were made in a field of @xmath32 t , which exceeds the zero - temperature upper critical field @xmath33t,@xcite and , for monitoring the superconducting transition , in a 0.6-t field across a restricted temperature range around the relevant @xmath4 . in order to achieve an acceptable signal - to - noise ( s / n ) ratio ,
the sample was powdered to a grain size of the order of 10@xmath16 m and kept in the form of a loose powder , hence reducing the electrical contact between the grains .
nmr measurements at the above - mentioned external magnetic fields were made at different temperatures between 1.6 and 300k by employing a conventional phase - coherent spectrometer .
the @xmath20p spectra were obtained by fast fourier transformation ( fft ) of the spin - echo signals generated by @xmath34 rf pulses with a 200-@xmath16s delay between pulses ( the measured transverse relaxation time is @xmath35 of ca . 500 @xmath16s ) .
since the employed `` hard '' rf pulses ( @xmath36 @xmath16s ) excite non - selectively the entire resonance line , no difference between fixed- and swept - frequency lineshapes was observed .
the nuclear spin - lattice relaxation times @xmath37 were measured at the peak position of each nmr line by means of the inversion - recovery method , using spin - echo detection at variable delays .
the magnetic field was calibrated via @xmath38al nmr of elemental aluminum , whose gyromagnetic ratio and knight shift are known to high precision .
this data was subsequently used to calculate the @xmath1p nmr shifts .
in fig . [ fig : lines ] we show the evolution of the @xmath1p lines with varying temperature at @xmath39 t , both above and below @xmath4 .
similar results with respect to the line widths and -shifts were obtained for @xmath40 t ( see fig . [
fig : shifts_2_vs_7 t ] ) .
representative @xmath1p nmr line shapes in srpt@xmath0p at @xmath39 t and temperatures in the range 1.75 to 290 k. the enhanced line widths and asymmetry below 5k reflect the onset of the superconducting phase . ]
we focus first on the normal phase . as can be seen in fig .
[ fig : shifts ] , the line width , of the order of 7khz at room temperature , increases gradually by approximately 50% down to ca .
12k and subsequently it stays constant between 12k and @xmath4 .
as for the line shifts in the normal phase , a moderate increase upon lowering the temperature is observed .
this increase is mostly linear down to approximately 50k , below which the line shift remains constant down to @xmath4 ( see inset of fig .
[ fig : shifts_2_vs_7 t ] ) .
the constancy of the shift above the superconducting transition is a feature often observed in simple metals with an only weakly - varying electronic susceptibility @xmath41 .
the relative shift , of the order of 0.12% , is close to that observed in simple metals such as ag and na and is the same for the 2- and 7-t external fields , respectively ( see fig . [
fig : shifts_2_vs_7 t ] ) . the line widths measured in a field of 7 t , instead , are approximately twice as large as those recorded at @xmath42 t , which , in turn , are three times larger than those at 0.6 t .
thus the linewidth scales almost linearly with the applied field .
the most interesting results are obtained below @xmath4 , where the inhomogeneous field distribution in the mixed state of a type - ii superconductor significantly broadens the nmr line .
[ fig : shifts ] shows the steep increase of the line width , from 10 to ca .
18 khz , between @xmath4 and 3.5k ( and from 3.2 to 8khz at 0.6 t - data not shown ) . upon a further reduction of @xmath29 ,
the width remains constant down to the lowest reached temperatures . at the same time the @xmath20p knight shift exhibits a sudden decrease below @xmath4 , a strong indication of spin - singlet superconductivity , since the pairing of electrons with opposite spins occurring in type @xmath5- ( or @xmath43-wave ) superconductors implies a significant decrease of the local spin susceptibility , as detected by the probe nuclei .
of the two possibilities , the experimental data rule out the occurrence of a @xmath43-wave sc pairing , since in that case , the drop of the knight shift would be less abrupt and the line broadening would reach its asymptotic @xmath44k value only gradually.@xcite from these static nmr results , an @xmath5-type pairing seems to be the most plausible configuration adopted by electrons in the superconducting state of srpt@xmath0p .
@xmath1p nmr shift ( left scale ) and linewidth ( right scale ) vs. temperature , as measured at an applied field of 2 t .
the drop in shift and the increase in width below @xmath4 are followed by a constant behavior at intermediate temperatures and by a final decrease at higher @xmath29 ( see text for details ) .
the increase of data scattering below @xmath4 reflects the decreased s / n ratio in the superconducting phase . ]
the conclusion about the @xmath5-wave pairing configuration is also supported by the @xmath20p spin - lattice @xmath45 relaxation - rate data displayed in fig .
[ fig : relaxations ] . the @xmath3 vs.
@xmath29 plot exhibits what is considered to be a typical signature of conventional bcs superconductivity , an exponential decrease with decreasing @xmath29 well within the superconducting phase , but no clear indication of a hebel - slichter - type coherence peak , which is masked by a distinct increase of @xmath3 in a narrow regime just above @xmath4 .
@xmath1p nmr spin - lattice relaxation data of srpt@xmath0p at selected temperatures measured at 2 t .
all fit curves exhibit a simple exponential ( @xmath46 ) recovery of the magnetization ( see text ) . ]
the spin - lattice relaxation times @xmath37 were evaluated from magnetization - recovery curves such as those shown in fig .
[ fig : t1_curves_2 t ] .
the nuclear magnetization data @xmath47 were fitted by using the single - exponential equation valid for spin-1/2 nuclei : @xmath48.\ ] ] here @xmath49 represents the saturation value of magnetization at thermal equilibrium , @xmath50 is the inversion factor ( exactly 2 for a complete inversion ) , and @xmath51 is a stretching exponent which accounts for possible distributions of relaxation rates.@xcite except at the lowest temperatures , the @xmath51 values were always found to be close to 1 , with no difference in fit quality whether @xmath51 was fixed or set free . considering the purely magnetic relaxation of @xmath52 nuclear probes , this implies the same local environment for all the @xmath20p nuclei , i.e. , disorder of either structural or magnetic nature is insignificant in our case .
temperature dependence of @xmath53 measured in a magnetic field of 2 t .
a similar cusp - like feature at @xmath4 is seen also at 0.6 t .
the decrease below the maximum occurring at @xmath4 is compatible with a fully - gapped superconductor ( see text ) .
inset : @xmath53 data over the full temperature range show a broad maximum at @xmath54k ( see also fig .
[ fig : suscept ] ) . ] up to @xmath55k , the normal - state spin - lattice relaxation rate varies approximately linearly with temperature ( see inset in fig . [
fig : relaxations ] ) , as is usually observed in conventional simple metals .
the progressive deviation from the linear - in-@xmath29 behavior above @xmath56 reflects a decrease in the dynamical spin susceptibility , as discussed in sec .
[ ssec : korringa ] and [ ssec : susceptibility ] .
focusing back on the superconducting state , just below @xmath4 we do not observe a distinct hebel - slichter ( coherence ) peak , but only a cusp - like feature peaking at @xmath4 .
analogous data taken in a 0.6-t field showed essentially the same type of feature in the vicinity of @xmath4 .
we recall that the hebel - slichter anomaly is due to pair coherence and to a quasiparticle density - of - states anomaly at the gap edge and is more pronounced at lower fields , where the pair - breaking effect of the field is less important.@xcite both these effects can be weakened by different causes , including interaction anisotropies and the broadening of the quasiparticle states due to strong - coupling effects .
indeed , while _ weakly_-coupled conventional @xmath5-wave superconductors , typically exhibit a hebel - slichter peak , this is not the case in the _ strong_-coupling limit , where the enhanced electron - phonon coupling implies reduced quasiparticle lifetimes.@xcite since the lack of a coherence peak can not rule out the possibility of an @xmath5-type electron pairing,@xcite further analysis is required .
the temperature dependence of @xmath37 at @xmath2 can provide information about the symmetry of the superconducting gap . in case of an anisotropic gap with nodes , a power - law dependence , @xmath57 ,
is expected since very low - energy electronic excitations around gap nodes can still contribute to relaxation .
on the other hand , for standard fully - gapped superconductors with no or a weak gap anisotropy , the relaxation rate depends exponentially on temperature , @xmath58 , with @xmath59 the superconducting gap at @xmath44 . in either case , non - negligible @xmath3 residual values
may be observed at the lowest temperatures .
is likely of intrinsic origin .
] most likely these are due to the presence of vortex - core or of thermally - excited vortex - motion relaxation mechanisms,@xcite which can be accounted for by an additional constant term .
while both models provide reasonable fits to the data , the power - law expression shows a somehow poorer fit quality and , most importantly , the resulting exponent varies significantly ( from 3.6 for the 2-t dataset to 2.7 for the data collected at 0.6 t ) . on the other hand , as shown in fig .
[ fig : relaxations ] , an exponential curve with @xmath60mev fits the srpt@xmath0p relaxation data satisfactorily well .
also the data taken at 0.6 t can be consistently fitted with a similar gap value , @xmath61mev , thus providing a good indication that the superconducting gap of srpt@xmath0p is nodeless . the value of the gap parameter , although slightly modified by the presence of the constant term , is in fair agreement with the 1.55mev value reported in ref . , but distinctly lower than @xmath11mev reported in ref . .
in any case , since the gap value from our analysis is clearly higher than 1.27 mev ( the gap expected from the bcs formula @xmath62 ) , it is compatible with a strong - coupling scenario .
besides the vortex - core relaxation mechanism mentioned above , the almost constant value of @xmath3 well below @xmath4 might also reflect a certain degree of superconducting - gap anisotropy , but we consider this as a less likely possibility .
some insight into the degree of electronic correlations in the normal state of a material can be gained by considering the so - called korringa relation@xcite @xmath63 here @xmath64 and @xmath65 are the electronic and the nuclear gyromagnetic ratios , respectively , while @xmath66 is the line shift due to the polarized conduction electrons .
( [ eq : korringa_r ] ) reflects the fact that both the readily accessible experimental parameters , the knight shift @xmath66 and the spin - lattice relaxation time @xmath37 , depend on the same electron - nucleus hyperfine interaction ( assumed to be mostly of fermi - contact character ) .
however , while for a simple ( fermi - gas ) metal the parameter @xmath67 is a constant ( depending only on @xmath65 of the probe - nucleus ) , for real materials the generalized relation@xcite @xmath68 , with @xmath69 a measure of the strength of the ( fermi - liquid ) quasiparticle interactions,@xcite is more appropriate .
since @xmath3 probes the dynamical susceptibility averaged over the brillouin zone , either fm or afm spin correlations will enhance it , but only fm correlations can significantly enhance also the shift .
thus , @xmath70 ( @xmath71 ) indicates the presence of ferro- ( antiferro-)magnetic electronic correlations . from the almost constant value of @xmath72 vs.
@xmath29 ( see inset in fig . [
fig : relaxations ] ) and considering the negligible orbital shift contribution ( see below ) , we find @xmath73 and , hence , @xmath74 in our case .
the resulting value , slightly larger than unity , seems to suggest the presence of ferromagnetic correlations in the normal state . however , the ratio @xmath73 is similar to 0.58 , 053 , and 0.50 observed in typical metals such as lithium , cesium , and silver , respectively.@xcite in these nearly - free - electron metals the moriya theory of exchange enhancement , whereby the exchange fluctuations also enhance the rate of the @xmath37 process,@xcite explains the experimental data satisfactorily .
consequently , we conclude that also in srpt@xmath0p the conduction electrons experience rather weak correlation effects , in good agreement with the results reported in ref . .
korringa relation [ reported in @xmath67 units , see eq .
( [ eq : korringa_r ] ) ] shows only a weak temperature dependence , with no distinct features at @xmath54k ( see also fig .
[ fig : suscept ] ) . ]
the magnetization @xmath75 was measured in an applied magnetic field of 2 t in specimens from the same batch as that used for nmr experiments .
the resulting values of the susceptibility , @xmath76 , for temperatures above 10k are shown in the main panel of fig .
[ fig : suscept ] .
the measured net susceptibility seems unusually small for a metal , because the contribution from the conduction electrons is almost exactly canceled by the large core diamagnetism mainly from pt ( @xmath77emu / mol per formula unit , according to standard tables ) . this compensation effect makes it possible to measure even subtle variations of the electronic contribution with @xmath29 . in order to relate @xmath23 with the nmr data , the relative @xmath1p nmr shifts in the two chosen magnetic fields ( 2 and 7 t , respectively )
are shown in fig .
[ fig : shifts_2_vs_7 t ] .
temperature dependence of the srpt@xmath0p molar magnetic susceptibility @xmath78 in an applied field of 2 t .
data were corrected by the addition of @xmath77emu / mol to account for the core diamagnetism ( see text ) .
inset : @xmath1p clogston - jaccarino ( @xmath79-@xmath80 ) plot at @xmath81 t .
the line is a fit to the @xmath82k data . ]
as expected for simple metals , the shifts are independent of the applied field and hence coincide perfectly .
below @xmath4 the high - field ( 7-t ) data reflect the normal phase : both the line shift and width remain practically constant upon decreasing temperature .
comparison of @xmath1p nmr line shifts vs. temperature at 2 ( circles ) and 7 t ( squares ) on logarithmic ( main panel ) and on linear scale ( inset ) .
the nature of the shift changes from constant to linearly decreasing at 45 k ( arrows ) .
the dashed lines highlight both regimes and serve as guides to the eye . ] from the @xmath23 data ( shown in fig .
[ fig : suscept ] ) and the nmr shifts ( see inset of fig . [
fig : shifts_2_vs_7 t ] ) , one recognizes immediately very similar features . in either case ,
a practically linear increase upon lowering the temperature , from 300k down to ca .
@xmath55k , is followed by a @xmath29-independent part .
this clear break in the temperature dependence of both @xmath83 and @xmath23 coincides with a maximum of the hall constant @xmath84 reported in ref . .
the reason for this might be that the trend to @xmath85 starts to dominate the @xmath86 contribution .
such a comparison of micro- ( nmr ) vs. macroscopic ( magnetometry ) results provides further information about the normal - state electronic properties of srpt@xmath0p . in our case ,
given the relative insensitivity to the applied field ( see fig .
[ fig : shifts_2_vs_7 t ] ) , it makes sense to compare the nmr shifts with the susceptibility data , as shown in the clogston - jaccarino@xcite ( or @xmath79-@xmath80 ) plot in the inset of fig .
[ fig : suscept ] . here
both data sets refer to the same ( 2-t ) field , with the temperature as an implicit parameter . from the slope @xmath87 of the @xmath79-@xmath80 plot for @xmath82k , and by assuming a standard electronic @xmath88-factor of 2.0 , the relation @xmath89 gives the hyperfine coupling constant , @xmath90t/@xmath91 , while the zero intercept with the @xmath79-axis gives the orbital shift , @xmath92% .
the latter is perfectly compatible with known orbital - shift values for a light nucleus such as @xmath1p . clearly , the orbital shift is negligible in comparison with the dominant magnetic shift .
we recall that hyperfine coupling constants are known to be strongly material - dependent , so it is not unusual that our value is an order of magnitude higher than , e.g. , @xmath93t/@xmath91 found in another phosphorus - based compound , pb@xmath94(vo)(po@xmath95)@xmath94.@xcite
below we discuss in more detail some aspects of the results presented in the previous sections . _ line shapes .
_ for all the applied magnetic fields the normal - state @xmath1p lines show a slight asymmetry , usually attributed to the crystallite surfaces or to minor impurities
. however , the persistence of the line asymmetry even in samples of different quality , most likely indicates a ( uniaxial ) anisotropy of the hyperfine coupling tensor ( whose study is beyond the scope of this work ) , which can be observed also in powdered samples.@xcite the uniaxial nature of the hf tensor is to be expected , if the symmetry of the distorted pt octahedra probed by the phosphorus nuclei is considered ( see structural unit in the inset of fig .
[ fig : magnetization ] ) .
the line shapes , in either the normal or the superconducting phase , indicate also the structural and chemical homogeneity of the sample .
an essentially single line with a relatively narrow width of only @xmath96khz ( see fig . [
fig : lines ] ) is typical of defect - free crystal structures , especially when considering the random polycrystalline nature of the sample .
similarly , the line asymmetry observed below @xmath4 , again reflects the high quality of the sample , since in a powdered superconductor with a significant degree of disorder the lines would appear symmetric . _ residual shift .
_ a key prediction of the bcs theory for superconductors with singlet @xmath5-wave pairing is the vanishing of the knight shift for @xmath97 , reflecting the vanishing susceptibility due to the pairing of quasiparticles with antiparallel spins . yet , even for elemental superconductors ( as , e.g. , sn or hg ) , a residual shift is observed even close to @xmath44.@xcite the presence of a significant orbital knight shift and/or of spin - orbit scattering ( known as ferrell - anderson effect ) have been identified as responsible for its occurrence .
the srpt@xmath0p case , however , is a bit more complex , since here @xmath20p is a low-@xmath98 nucleus which interacts with superconducting carriers arising mostly from in - plane @xmath13-hybridized states between pt and p ions.@xcite since orbital shifts are negligible for the light probe - nuclei , the presence of a low - temperature nonzero knight shift in srpt@xmath0p is , therefore , most likely due to spin - orbit scattering effects . _
core - polarization effects and @xmath56 anomaly .
_ another contribution to the nonvanishing shift at @xmath44 can arise from core - polarization effects,@xcite whereby the exchange interaction of localized core electrons with field - polarized conduction band electrons ( partially of @xmath43 character in our case ) , implies an additional nonzero contact interaction in the @xmath20p nuclei .
the observation of large negative @xmath99pt - nmr knight shifts ( data not shown ) seems to confirm the presence of core - polarization effects in srpt@xmath0p.@xcite while this contribution to the knight shift would disappear entirely if purely @xmath43-band electrons were coupled into cooper pairs , a residual shift would still arise considering the hybridized ( i.e. , not exclusively @xmath43 ) character of the conduction band quasiparticles .
the consequences of core - polarization effects are more important , however , in the normal state .
the electronic contribution to the magnetic susceptibility of a @xmath43-type band is generally temperature dependent , thus implying a temperature - dependent knight shift , as actually observed ( for an ideal metal one would have expected a constant shift ) .
while core - polarization effects might explain the similar @xmath29-dependence of @xmath79 and@xmath100 , the abrupt change in slope observed at @xmath56 , as well as the value of @xmath56 itself , remain to be investigated .
_ weak electron correlation .
_ our nmr data of srpt@xmath0p , indicate a fairly simple metallic behavior in the normal state , followed by standard bcs features in the superconducting state .
these results are in remarkable agreement with the transport and magnetometry measurements reported in ref . in two important aspects .
firstly , both nmr ( via the korringa product ) and the macroscopic measurements ( via the sommerfeld - wilson ratio ) demonstrate the _ absence of significant electron correlations_. secondly , there is a distinct feature at @xmath55k in the @xmath20p knight shift and in the @xmath3 relaxation behavior , respectively , both of which change from a constant - in-@xmath29 to a linear - in-@xmath29 decrease as the temperature changes across @xmath56 .
this temperature coincides with the above - mentioned maximum in @xmath101 , the latter reflecting anisotropies of the electron mean free path across different parts of the fermi surface .
the latter fact is consistent with conclusions drawn in ref . .
the coincidence of @xmath56 values , derived from results of very different types of experiments , is unclear at present . as for the korringa relation ,
we recall that the absolute value of @xmath69 might also be influenced by extraneous factors , such as the presence of strong disorder,@xcite or effects related to the hyperfine form factor ( i.e. , @xmath102-space filtering ) .
nevertheless , as mentioned above , these can be excluded in our case and the observation of a constant korringa ratio vs. temperature represents a valid proof of fermi - liquid behavior in srpt@xmath0p , where @xmath73 is constant over a fairly wide temperature range . _
centro- vs. noncentro - symmetric superconductors .
_ we conclude with a comparison of srpt@xmath0p with two noncentrosymmetric superconductors , the non - magnetic lapt@xmath0si ( @xmath103k ) and the magnetic cept@xmath0si ( @xmath104k ) , both of which were studied via nmr , as well.@xcite the latter not only share the same structure , but show also similar antisymmetric spin - orbit couplings .
hence , a comparison of the three is relevant to understand whether it is the lack of inversion symmetry , or the presence of rare - earth ions to account for the unusual properties of noncentrosymmetric superconductors . while srpt@xmath0p and lapt@xmath0si both
follow a perfect korringa law down to @xmath4 , in cept@xmath0si the @xmath3 ratio is enhanced upon cooling due to the development of 4@xmath50-derived magnetic fluctuations .
most importantly , unlike in srpt@xmath0p and lapt@xmath0si , where the superconducting transition is of conventional type and not dominated by correlation effects , in cept@xmath0si , the superconductivity emerges from a unique heavy - electron state which coexists with antiferromagnetic order . from these facts
, we conclude that the key differences between the considered centro- and non - centrosymmetric superconductors , are most likely related to the presence of ce@xmath105 ions and the ensuing heavy - fermion phenomena , while the parity mixing alone ( common to both lapt@xmath0si and cept@xmath0si ) can not explain the peculiar behavior of the latter . despite the many similarities between srpt@xmath0p and lapt@xmath0si , their belonging to different classes of crystal symmetry
may possibly explain their distinctly different critical temperatures .
incidentally , from the available data in literature , also the centrosymmetric lapt@xmath0p compound seems most likely to be a conventional bcs type - i superconductor .
we presented the results of @xmath1p nmr measurements in the recently discovered srpt@xmath0p superconducting compound . in the normal state ,
the nuclear spin - relaxation rate obeys a korringa - type relation @xmath106 ( sk)@xmath107 and the line shift is independent of temperature . in the superconducting state ,
the lack of a clear coherence peak in @xmath108 just below @xmath4 , most likely due to a strong electron - phonon coupling , is followed by an exponential , thermally - activated decrease upon further cooling .
a similar behavior is shown also by the line shift .
the overall results of our nmr investigation make us conclude that srpt@xmath0p , which behaves as a standard metal in the normal state , is characterized by singlet pairing of the electrons , and a gap function with conventional @xmath5-wave - type symmetry .
finally , a comparison with the analogous non - centrosymmetric superconductors lapt@xmath0si and cept@xmath0si , strongly suggests that the peculiarities of the latter are closely related to the presence of rare - earth cations .
the lack of inversion symmetry may cause a much lower @xmath4 , however .
the authors thank r. khasanov ( paul scherrer institut ) for sharing his data prior to publication and for useful discussions . this work was financially supported in part by
the schweizerische nationalfonds zur frderung der wissenschaftlichen forschung ( snf ) and the nccr research pool manep of snf .
99 t. takayama , k. kuwano , d. hirai , y. katsura , a. yamamoto , and h. takagi , `` strong coupling superconductivity at 8.4k in an antiperovskite phosphide srpt@xmath0p , '' phys .
lett . * 108 * , 237001 ( 2012 ) .
e. bauer , r. lackner , g. hilscher , h. michor , m. sieberer , a. eichler , a. gribanov , y. seropegin , and p. rogl , `` rept@xmath0si ( re = la , pr , nd , sm and gd ) : isotypes of the heavy fermion superconductor cept@xmath0si , '' j. phys . : condens .
matter * 17 * , 1877 ( 2005 ) .
m. yogi , y. kitaoka , s. hashimoto , t. yasuda , r. settai , t. d. matsuda , y. haga , y. nuki , p. rogl , and e. bauer , `` evidence for a novel state of superconductivity in noncentrosymmetric cept@xmath0si : a @xmath109pt - nmr study , '' phys .
lett . * 93 * , 027003 ( 2004 ) .
m. yogi , y. kitaoka , s. hashimoto , t. yasuda , r. settai , t. d. matsuda , y. haga , y. nuki , p. rogl , and e. bauer , `` @xmath109pt - nmr study on noncentrosymmetric heavy - fermion superconductor cept@xmath0si , '' j. phys .
solids * 67 * , 522 ( 2006 ) .
h. mukuda et al . , `` multiband superconductivity in heavy - fermion compound cept@xmath0si without inversion symmetry : an nmr study on a high - quality single crystal , '' j. phys .
jpn . * 78 * , 014705 ( 2009 ) .
e. bauer , m. sigrist , eds .
, _ non - centrosymmetric superconductors _ , lecture notes in physics , vol .
847 ( springer , berlin , 2012 ) .
h. chen , x .- f .
xu , c. cao , and j. dai , `` first - principles calculations of the electronic and phonon properties of @xmath110pt@xmath111p ( @xmath110 @xmath112 ca , sr , and la ) : evidence for a charge - density - wave instability and a soft phonon , '' phys .
b * 86 * , 125116 ( 2012 ) .
a. subedi , l. ortenzi , and l. boeri , `` electron - phonon superconductivity in @xmath110pt@xmath111p ( @xmath113 , ca , la ) compounds : from weak to strong coupling , '' phys .
b * 87 * , 144504 ( 2013 ) .
kang , k .- h .
ahn , k .- w .
lee , and b. i. min , `` electron and phonon band - structure calculations for the antipolar srpt@xmath0p antiperovskite superconductor : evidence of low - energy two - dimensional phonons , '' j. phys .
. jpn . * 82 * , 053703 ( 2013 ) .
r. khasanov , a. amato , p. k. biswas , h. luetkens , n. d. zhigadlo , and b. batlogg , `` srpt@xmath0p : a two - band single - gap superconductor , '' phys .
b * 90 * , 140507(r ) ( 2014 ) .
r. e. walstedt , _ the nmr probe of high-@xmath4 materials _ ( springer verlag , berlin , 2008 )
. n. j. curro , `` nuclear magnetic resonance in the heavy fermion superconductors , '' rep .
* 72 * , 026502 ( 2009 ) .
t. shiroka , f. casola , v. glazkov , a. zheludev , k. pra , h .- r .
ott , and j. mesot , `` distribution of nmr relaxations in a random heisenberg chain , '' phys .
lett . * 106 * , 137202 ( 2011 ) . y. masuda and n. okubo , `` nuclear spin - lattice relaxation in superconducting mixed state , '' j. phys
. soc . jpn .
* 26 * , 309 ( 1969 ) .
ohsugi , y. kitaoka , m. kyogaku , k. ishida , k. asayama , and t. ohtani , `` nuclear relaxation study in strong coupling superconductors a comparison with high-@xmath4 superconductors , '' j. phys .
. jpn . * 61 * , 3054 ( 1992 ) .
d. parker , o. v. dolgov , m. m. korshunov , a. a. golubov , i. i. mazin , `` extended @xmath114 scenario for the nuclear spin - lattice relaxation rate in superconducting pnictides , '' phys .
b * 78 * , 134524 ( 2008 ) .
d. e. maclaughlin , magnetic resonance in the superconducting state , in _ solid state physics _ ,
31 , h. ehrenreich , f. seitz , and d. turnbull , eds . , ( academic press , new york , 1976 ) , pp .
169 . m. corti , b. j. suh , f. tabak , a. rigamonti , f. borsa , m. xu , and b. dabrowski , `` flux - line dynamics in yba@xmath94cu@xmath95o@xmath115 from @xmath116y nmr , '' phys .
b * 54 * , 9469 ( 1996 ) . j. korringa , `` nuclear magnetic relaxation and resonance line shift in metals , '' physica * 16 * , 601 ( 1950 ) .
a. narath and h. t. weaver , `` effects of electron - electron interactions on nuclear spin - lattice relaxation rates and knight shifts in alkali and noble metals , '' phys
. rev . * 175 * , 373 ( 1968 ) .
d. pines , `` paramagnetic susceptibility of conduction electrons , '' phys . rev . * 95 * , 1090 , ( 1954 ) . t. moriya , `` the effect of electron - electron interaction on the nuclear spin relaxation in metals , '' j. phys . soc .
* 18 * , 516 ( 1963 ) .
a. m. clogston , v. jaccarino , y. yafet , `` interpretation of knight shifts and susceptibilities of transition metals : platinum , '' phys . rev . * 134 * , a650 ( 1964 ) .
r. nath , y. furukawa , f. borsa , e. e. kaul , m. baenitz , c. geibel , and d. c. johnston , `` single - crystal @xmath20p nmr studies of the frustrated square - lattice compound pb@xmath94vo(po@xmath95)@xmath94 , '' phys .
b * 80 * , 214430 ( 2009 ) .
a. kawamoto , k. miyagawa , y. nakazawa , k. kanoda , `` @xmath117c nmr study of layered organic superconductors based on bedt - ttf molecules , '' phys .
* 74 * , 3455 ( 1995 ) .
a. j. freeman and r. e. watson , hyperfine interactions in magnetic materials , in _ magnetism _ , vol .
2a , g. t. rado and h. suhl , eds .
, ( academic press , new york , 1965 ) , pp 168305 . b. s. shastry and e. abrahams , `` what does the korringa ratio measure ? , '' phys .
lett . * 72 * , 1933 ( 1994 ) . | we report a study of the normal- and superconducting - state electronic properties of the centrosymmetric compound srpt@xmath0p via @xmath1p nuclear - magnetic - resonance ( nmr ) and magnetometry investigations .
essential features such as a sharp drop of the knight shift at @xmath2 and an exponential decrease of the nmr spin - lattice relaxation ratio @xmath3 below @xmath4 are consistent with an @xmath5-wave electron pairing in srpt@xmath0p , although a direct confirmation in the form of a hebel - slichter - type peak is lacking .
normal - state nmr data at @xmath6k indicate conventional features of the conduction electrons , typical of simple metals such as lithium or silver .
our data are finally compared with available nmr results for the noncentrosymmetric superconductors lapt@xmath0si and cept@xmath0si , which adopt similar crystal structures . |
for the general case of @xmath1 transition into a combination of @xmath27 ( @xmath28 ) and @xmath29 ( @xmath30 ) states one can write the sk , sno cc and sno nc rates as r^el_sk & = & f_b p_ee + f_b r p_ea , [ two ] + r^cc_sno & = & f_b p_ee , [ three ] + r^nc_sno & = & f_b ( p_ee + p_ea ) , [ four ] where @xmath16 and @xmath31 denote the probabilities folded with the detector response function @xcite and averaged over energy . to extract a model independent bound on @xmath16 one has to ensure an equality of the response functions which amounts to slight adjustment of the sk threshold energy and the rate @xcite .
our approach is slightly different .
we treat @xmath16 to be effectively energy independent .
the sk spectrum data indicates a flat probability down to 5 mev @xcite .
this is corroborated by sno @xcite which now has a threshold of 5 mev for kinetic energy of the observed electron .
hence we consider this assumption as justified and expect the results to be insensitive to the differences in the response functions .
it should be noted here that in cotrast to the sno cc events their nc events correspond to a neutrino energy threshold of 2.2 mev .
however it is clear from eq .
( 5 ) that for a @xmath1 to @xmath28 transition there is no reason to expect any energy dependence in @xmath32 .
on the other hand for the general case of @xmath1 transition into a combination of @xmath28 and @xmath30 our approach effectively assumes @xmath33 to be energy independent down to 2.2 mev .
a comparison of the current values @xmath19 with @xmath32 is shown in fig .
1 . it constitutes a 5.3@xmath34 signal for transition to a state containing an active neutrino component or alternatively a 5.3@xmath34 signal against a pure sterile solution .
next we consider the general case where @xmath1 goes to a mixed state = @xmath35
. then one can write @xmath36 ) . substituting this in the equations ( [ two ] ) and ( [ four ] ) and eliminating @xmath16 using equation ( [ three ] )
one gets the following set of equations for @xmath24 and @xmath37 @xcite ^2(f_b - r^cc_sno ) & = & ( r^el_sk - r^cc_snc)/r , [ five ] + ^2(f_b - r^cc_sno ) & = & r^nc_sno - r^cc_sno .
[ six ] we treat @xmath37 as a model parameter . and for different input values of @xmath37 we determine the central value and the @xmath38 and @xmath39 ranges of @xmath24 by taking a weighted average of the equations ( [ five ] ) and ( [ six ] ) .
the corresponding curves are presented in fig .
2 . combining the 2 @xmath34 lower limit of @xmath24 from this fit with the @xmath39 upper limit from the ssm ( vertical lines )
gives a lower limit of @xmath37 @xmath40 0.45 i.e. the probability of the active component is @xmath40 45% .
note that there is no upper limit on this quantity - i.e. the data is perfectly compatible with @xmath1 transition into purely active neutrinos . assuming transition into purely active neutrinos ( @xmath41 ) we show in fig .
3 the @xmath38 and @xmath39 contours in the @xmath42 plane from the combinations sk+snocc and sk+snocc+snonc .
the inclusion of the nc rate narrows down the ranges of @xmath24 and @xmath16 .
the error in @xmath24 after the inclusion of nc data is about half the size of the corresponding error from ssm as is seen from fig .
in this section we present the results of our @xmath43 analysis of solar neutrino rates and sk spectrum data in the framework of two flavour oscillation of @xmath1 to an active flavour .
we use the standard techniques described in our earlier papers @xcite excepting for the fact that instead of the quantities @xmath44 and @xmath19 we now fit the ratios @xmath20 and @xmath21 .
the @xmath0 flux normalisation gets cancelled from these ratios and the analysis becomes independent of the large ( 16 - 20% ) ssm uncertainty associated with this .
we include in our global analyses the 1496 day sk zenith angle spectra @xcite .
since we use both sk rate and sk spectrum data we keep a free normalisation factor for the sk spectrum .
this amounts to taking the information on total rates from the sk rates data and the information of the spectral shape from the sk spectrum data .
the sno cc and nc rates have a large anticorrelation .
we have taken into account this correlation between the measured sno rates in our global analyses .
further details of this fitting method can be found in @xcite . in table 2
we present the best - fit parameters , @xmath45 and goodness of fit ( gof ) .
the best - fit comes in the high(lma ) region as before @xcite .
however as is seen from fig .
4a the incorporation of the nc data narrows down the allowed regions , and in particular the low region becomes much smaller .
we have also performed an alternative @xmath43 fit to the rates of table 1 @xcite along with the 1496 day sk spectra @xcite , keeping @xmath24 as a free parameter . even though we allow @xmath24 to vary freely the nc data serves to control @xmath24 within a range determined by its error . as we see from table 2 and fig . 4b
the results of this fit are very similar to the previous case .
the best fit comes from the high(lma ) region , while no allowed region is obtained for the low solution at the 99% cl level .
maximal mixing is seen to be disallowed at the @xmath3 level . to illustrate the impact of the nc rate on the oscillation solutions we have repeated the free @xmath24 fit without this rate .
the results are shown in fig 4c .
evidently the nc data plays a pivotal role in constraining the oscillation solutions , particularly in the low / qvo region , which is allowed only at the @xmath3 level .
it puts an upper bound on the @xmath46 in the lma region and rules out maximal mixing .
the first sno nc data constitutes a 5.3@xmath34 signal for transition into a state containing an active neutrino component .
the inclusion of this data puts much tighter constraints on @xmath24 and @xmath16 from a model independent analysis involving active neutrinos as compared to the sno cc / sk combination . in this paper
we have discussed two useful strategies , of incorporating the nc data in the global @xmath43 analysis of rates and spectrum data , by which one can avoid the large @xmath0 flux uncertainty from the ssm .
+ @xmath47 we fit the ratios of the sk elastic and sno cc rates w.r.t the nc rate , from which the @xmath24 cancels out .
+ @xmath47 we fit the rates by keeping @xmath24 as a free parameter , where the inclusion of the sno nc rate (= @xmath24 ) serves to control this parameter .
both the analyses give very similar results .
they clearly favour the high(lma ) solution , while a limited region of the low solution is also acceptable at the @xmath3 level .
the maximal mixing solution is disfavoured at the @xmath3 level . as more data
accumulate one expects a substantial reduction in the error bar of the sno nc rate , resulting in further tightening of the allowed regions of neutrino mass and mixing .
+ + _ note added _ : the paper @xcite appeared on the net after completion of our work . in the region of overlap
our results agree with theirs as well as with the updated version of @xcite . it may be added here that the sno cc and nc rates given in table 1 are obtained assuming undistorted energy spectra above 5 mev , which for transitions to active neutrinos has good empirical justification as mentioned above .
we thank prof .
mark chen of sno collaboration for communication on this point .
a. bandyopadhyay , s. choubey , s. goswami and k. kar , phys . lett . *
b519 * , 83 ( 2001 ) ; s. choubey , s. goswami , k. kar , h.m . antia and s.m .
chitre , phys . rev . *
d64 * , 113001 ( 2001 ) ; s. choubey , s. goswami and d.p .
roy , phys .
rev . * d65 * , 073001 ( 2002 ) ; a. bandyopadhyay , s. choubey , s. goswami and k. kar , phys . rev . * d65 * , 073031 ( 2002 ) .
fogli , e. lisi , d. montanino , a. palazzo , phys . rev . *
d64 * , 093007 ( 2001 ) ; j.n .
bahcall , m.c .
gonzalez - garcia , c. pana - garay , jhep * 0108 * , 014 ( 2001 ) ; p.i .
krastev and a.yu .
smirnov , e - print archive : hep - ph/0108177 ; m.v .
garzelli and c. giunti , jhep * 0112 * , 017 ( 2001 ) .
j. n. abduratshitov et al . ,
sage collaboration , astro - ph/0204245 ; w. hampel _
et al . , _ gallex collaboration , phys . lett . *
b447 * , 127 ( 1999 ) ; m. altman _
et al . , _ gno collaboration , phys . lett . *
b490 * , 16 ( 2000 ) ; e. belloti , talk at gran sasso national laboratories , may 17 , 2002 ; t. kirsten , talk at neutrino 2002 , munich . .
the observed solar neutrino rates relative to the @xmath49 predictions ( bp2000 ) are shown along with their compositions for different experiments . for the @xmath50 experiment
the @xmath1 contribution to the rate @xmath51 is shown in parantheses assuming @xmath52 transition . in the combined ga rate we have included the latest data from sage and gno . [ cols="^,^,^ " , ] | we perform model independent and model dependent analyses of solar neutrino data including the neutral current event rate from sno .
the inclusion of the first sno nc data in the model independent analysis determines the allowed ranges of @xmath0 flux normalisation and the @xmath1 survival probability more precisely than what was possible from the sk and sno cc combination .
we perform global @xmath2 oscillation analyses of solar neutrino data using the nc rate instead of the ssm prediction for the @xmath0 flux , in view of the large uncertainty in the latter .
the lma gives the best solution , while the low solution is allowed only at the @xmath3 level .
* implications of the first neutral current data from sno for solar neutrino oscillation * abhijit bandyopadhyay@xmath4 , sandhya choubey@xmath5 , srubabati goswami@xmath6 , d.p .
roy@xmath7 @xmath4saha institute of nuclear physics , bidhannagar , kolkata 700 064 , india + @xmath5 department of physics and astronomy , university of southampton , highfield , southampton s017 1bj , uk + @xmath8 harish - chandra research institute , chhatnag road , jhusi , allahabad - 211 - 019 , india @xmath9tata institute of fundamental research , homi bhabha road , mumbai 400 005 , india + @xmath10physics department , university of california , riverside , ca 92521,usa the neutral current results from the sudbury neutrino observatory measures for the first time the total flux of @xmath0 neutrinos coming from the sun @xcite . in a recent paper @xcite we had examined the role of the anticipated nc data from sno in enhancing our understanding of the solar neutrino problem .
the sno nc rate can be expressed in terms of sno cc and sk elastic scattering rates as @xcite r^nc_sno = r^cc_sno + ( r^el_sk - r^cc_sno)/r , [ one ] where @xmath11 for a threshold energy of 5 mev ( including the detector resolutions and the radiative corrections to @xmath12 scattering cross - sections ) .
all the rates are defined with respect to the bbp2000 standard solar model ( ssm ) @xcite .
we showed in @xcite that because sno has a greater sensitivity to the nc scattering rate as compared to sk , the sno nc measurement will be more precise and hence incorporation of this can be more predictive than the sno cc and sk combination .
we took three representative nc rates @xmath13 = 0.8,1.0 and 1.2 ( @xmath14 ) and showed that 1 . for a general transition of @xmath1 into a mixture of active and sterile neutrinos the size of the sterile component
can be better constrained than before .
2 . for transition to a purely active neutrino the @xmath15 neutrino flux normalisation and the survival probability @xmath16
are determined more precisely .
3 .
we had also performed global two flavour oscillation analysis of the solar neutrino data for the @xmath17 case , where instead of @xmath18 and @xmath19 we used the quantities @xmath20 and @xmath21 .
these ratios are independent of the @xmath22 flux normalisation and hence of the ssm uncertainty .
we showed that use of these ratios can result in drastic reduction of the allowed parameter regions specially in the low - qvo area depending on the value of the nc rate .
we now have the actual experimental result r^nc_sno = 1.01 0.12 while eq .
( [ one ] ) gives @xmath23 .
thus in 306 live days ( 577 days ) the sno nc measurement has achieved a precision , which is already better than that obtained from the sk and sno cc combination .
this paper follows closely the analysis that we have done in @xcite but incorporating the actual data .
in addition we also perform an alternative global analysis for @xmath17 oscillation by letting the @xmath0 normalisation factor @xmath24 vary freely , where the inclusion of @xmath25 in the fit serves to control this parameter .
as we shall see below the two methods of global analysis give very similar results .
in section 1 we discuss the constraints on the electron neutrino survival probability , the @xmath0 normalisation factor @xmath24 and the fraction of sterile component without assuming any particular model for the probabilities . in section 2
we perform the global analyses assuming two flavour @xmath26 oscillation . |
the flavor evolution of supernova ( sn ) neutrinos , is strongly impacted by the self - induced effects , associated with instabilities induced by the the neutrino - neutrino interactions in the deepest stellar regions @xcite ( see also @xcite for a recent review ) . in this context a key ingredient in the characterization of these effects is related to the current - current nature of low - energy weak interactions , which implies a `` multi - angle term '' @xcite @xmath4 , where @xmath5 is the neutrino velocity @xcite . till recently
, all studies have assumed the axial symmetry for this term , that would then depend only on the zenith - angle .
it has been shown that the `` multi - zenith - angle '' ( mza ) term in some cases can hinder the maintenance of the coherent oscillation behavior for different neutrino modes @xcite . a valuable tool to diagnose the possible instabilities of a dense neutrino gas is given by the linearized stability analysis of the neutrino equations of motion .
the linearized equations including generic azimuthal and zenith angle distributions for neutrinos were first worked out in @xcite . then , the formalism for the azimuthal symmetric case was further developed in @xcite .
the stability method would allow one to determine the possible onset of the flavor conversions , seeking for an exponentially growing solution of the eigenvalue problem , associated with the linearized equations of motion for the neutrino ensemble .
recently , this method has been applied in @xcite removing the assumption of axial symmetry in the @xmath6 propagation . as a result , a multi - azimuthal - angle ( maa )
instability was found in addition to the bimodal @xcite and mza ones @xcite . in particular , it was considered a neutrino ensemble with a strong excess of @xmath7 over @xmath8 , as expected during the sn accretion phase ( at post - bounce times @xmath9 s ) . in this situation , the instability has been found in normal mass hierarchy ( nh , @xmath10 ) , where the system would have been stable imposing the perfect axial symmetry .
subsequently , the role of this instability has been clarified in @xcite with simple toy models .
the discovery of the new maa effects , has also motivated first numerical studies of the non - linear neutrino propagation equations in sn , introducing the azimuthal angle as angular variable in the multi - angle kernel , in addition to the usual zenith angle @xcite .
remarkably it was considered the @xmath6 propagation only along a radial direction , i.e. a local solution along a specific line of sight , under the assumption that the transverse variations of the global solution are small . in this approximation , for the unstable case discussed above , maa effects lead in nh to spectral swaps and splits analogous to what produced in inverted hierarchy ( ih , @xmath11 ) by the known bimodal instability @xcite also in a completely isotropic neutrino gas @xcite .
all these recent works assume that the maa effects can develop without any matter hindrance .
remarkably , a crucial ingredient to be considered to determine the impact of self - induced flavor instabilities is the ordinary matter term , associated with the net electron densities @xmath12 in sne .
as pointed out in @xcite for the axial - symmetric case , when @xmath12 is not negligible with respect to the neutrino density @xmath13 , the large phase dispersion induced by the matter for @xmath6 s traveling in different directions , would partially or totally suppress the self - induced oscillations through peculiar mza effects .
recent studies of this case performed with realistic sn models , indicates that this situation is realized during the supernova accretion phase ( at post - bounce times @xmath9 s ) . as a consequence
, the self - induced flavor conversions found in ih in the axial - symmetric models are strongly inhibited @xcite . from a preliminary schematic study done in @xcite it results that the matter density required to suppress the mma instability in nh
is larger than the one necessary to suppress the self - induced conversions in ih for the axial symmetric case .
motivated by these previous results , we find it is mandatory to understand what is the role of the dense matter on the maa effects during the accretion phase . we will use as benchmark for the neutrino and matter density profiles the sn models from recent long term simulations of core - collapse explosions , performed by the basel - darmstadt model .
these were already considered by some of us in @xcite .
the plan of our work is as follows . in sec . 2 we present the neutrino equations of flavor evolution without imposing axial asymmetry .
then , we describe the setup to perform the stability analysis of the linearized equations of motion . in sec .
3 we present the results of the stability analysis and we compare them with the numerical solution of the equations .
we focus on the accretion phase for two different sn progenitor models . finally , in sec .
4 we comment on our results and we conclude .
in the axial symmetric case , the sn neutrino flavor evolution is described by ordinary differential equations @xcite , characterizing the flavor changes along a radial direction .
when axial symmetry is broken by the maa effects , in order to get the global solution of the problem in general one would consider also variations along the transverse direction to the neutrino propagation .
this would imply passing from ordinary to partial differential equations , with a big layer of complication in the numerical solution .
however , in our study we are mostly interested in answering the question of the stability of the dense neutrino gas under mma effects , in the presence of a large matter term .
therefore , _ before _ the maa instability emerges , we can still consider the ordinary differential equations in the only radial direction .
these are enough to determine which cases are completely stable under maa effects . in the other cases , in which the maa instability develops
, these equations would be useful to determine the onset radius of the flavor conversions .
nevertheless , the subsequent flavor evolution can be taken just as indicative , since it is based on the assumption that variations in the transverse direction always remain small . under this approximation
, the flavor evolution depends only on @xmath14 , @xmath15 and @xmath5 .
then , following @xcite we write the equations of motion for the flux matrices @xmath16 as function of the radial coordinate .
we use negative energy @xmath15 for anti - neutrinos . following the usual prescription
, we label the zenith angular mode in terms of the variable @xmath17 , where @xmath18 is the emission angle relative to the radial direction of the neutrinosphere radius @xmath19 @xcite .
we call @xmath20 the azimuth angle of the neutrino velocity @xmath21 .
we normalize the flux matrices to the @xmath22 number flux at the neutrinosphere .
the diagonal @xmath16 elements are the ordinary number fluxes integrated over a sphere of radius @xmath14 . the off - diagonal elements , which are initially zero , carry a phase information due to flavor mixing . then , the equations of motion read @xcite @xmath23 \,\ , \label{eq : eom1}\ ] ] with the hamiltonian @xcite @xmath24 the matrix @xmath25 of neutrino mass - squares causes vacuum flavor oscillations .
we work in a two - flavor scenario , associated with the atmospheric mass - square difference @xmath26 ev@xmath27 and a small ( matter suppressed ) in - medium mixing @xmath28 .
we will always consider nh , where maa effect could emerge for the flux ordering we are considering .
the matrix @xmath29 in flavor basis , contains the net electron density and is responsible for the mikheyev - smirnov - wolfenstein ( msw ) matter effect @xcite with the ordinary background .
finally , the term at second line represents the @xmath6-@xmath6 refractive term , where @xmath30 . in the multi - angle term of eq .
( [ eq : eom2 ] ) , the radial velocity of a mode with angular label @xmath31 is @xmath32 @xcite and the transverse velocity is @xmath33 @xcite .
the term @xmath34 is the responsible for the breaking of the axial symmetry . to solve numerically eq .
( [ eq : eom1 ] ) we use an integration routine for stiff ordinary differential equations taken from the nag libraries @xcite and based on an adaptive method .
we have used @xmath35 modes for @xmath36 $ ] , @xmath37 for @xmath38 $ ] .
concerning neutrino emission model , in order to simplify the complexity of our numerical simulations of the flavor evolution , we assume all @xmath6 s to be represented by a single energy , that we fix at @xmath39 mev . this approximation is reasonable since our main purpose is to determine only if the dense matter effects block the development of the self - induced transformations . in order to perform the stability analysis , we linearize the equations of motion [ eq .
( [ eq : eom2 ] ) ] , following the approach of @xcite .
we write the flux matrices in the form @xmath40 where we switch to the frequency variable @xmath41 , and we introduce the neutrino flux difference distributions @xmath42 that represent the flavor fluxes @xmath43 at the neutrinosphere , normalized to the @xmath22 flux . in the following
we will always assume axial symmetry of the neutrino emission .
therefore @xmath44 .
the @xmath45 is conserved and then irrelevant for the flavor conversions .
the @xmath7 survival probability is @xmath46 , given in terms of the swap factor @xmath47 of the matrix in the second term on the right - hand side .
the off - diagonal components @xmath48 are complex and @xmath49 .
the initial conditions are @xmath50 and @xmath51 .
self - induced flavor transitions start when the off - diagonal term @xmath48 grows exponentially . in the small - amplitude limit @xmath52 , and at far distances from the neutrinosphere @xmath53 , the linearized evolution equations for @xmath48 assume the form @xcite @xmath54 s \nonumber \\ & - & \mu \int d \gamma^\prime [ u + u^\prime - 2\sqrt{u u^\prime } \cos ( \varphi-\varphi^\prime)]g^\prime s^\prime \label{eq : lin } \,\ .\end{aligned}\ ] ] in this equation @xmath55 , quantifies the `` asymmetry '' of the neutrino spectrum , normalized to the @xmath22 number flux . in the sn models we are using ,
typically @xmath56 during the accretion phase ( see fig . 3 in @xcite ) .
the @xmath6-@xmath6 interaction strength is given by @xmath57}{4 \pi r^2}\frac{r_{\nu}^2}{2 r^2 } \nonumber \\ & = & \frac{3.5 \times 10^{9}}{r^4 } \left(\frac{l_{{\overline\nu}_e}}{\langle e_{{\overline\nu}_e } \rangle } - \frac{l_{{\overline\nu}_x}}{\langle e_{{\overline\nu}_x } \rangle } \right ) \nonumber \\ & \times & \left(\frac{15 \,\ \textrm{mev } } { 10^{52 } \,\
\textrm{erg / s}}\right ) \left(\frac{r_{\nu}}{10 \,\
\textrm{km } } \right)^2 , \label{eq : mu}\end{aligned}\ ] ] while ordinary matter background term is given by @xmath58 where @xmath59 is the net electron fraction , and @xmath60 is the matter density .
the radial distance @xmath14 is expressed in km , while the numerical values of @xmath61 and @xmath62 in the two previous equations are quoted in km@xmath63 , as appropriate for the sn case .
one can write the solution of the linear differential equation [ eq . ( [ eq : lin ] ) ] in the form @xmath64 with complex frequency @xmath65 and eigenvector @xmath66 . a solution with @xmath67
would indicate an exponential increase in @xmath48 , i.e. an instability .
the solution of eq .
( [ eq : lin ] ) can then be recast in the form of an eigenvalue equation for @xmath66 .
one gets as consistency condition @xcite @xmath68 a flavor instability is present whenever eq .
( [ consit ] ) admits a solution @xmath69 .
we consider the core - collapse supernova simulations of massive stars with 8.8 and 10.8 @xmath2 progenitor from ref .
@xcite , taken as benchmark for our numerical study in @xcite .
the first type of sn belongs to the class of o - ne - mg - core progenitor .
the second one is an iron - core progenitor . under the single - energy approximation we are using ,
we characterize the neutrino energy spectra as @xmath70 \,\ , \label{singleen } \end{aligned}\ ] ] where @xmath71 are the total number fluxes of the species @xmath72 at the neutrinosphere . in order to fix the neutrinosphere radius @xmath73 , consistently with our choice in @xcite
we take the radius at which the @xmath7 s angular distribution has no longer significant backward flux , i.e. a few % of the total one .
this typically is in the range @xmath74 km ( see fig . 4 in @xcite ) . for our choice of neutrino representative energy ( @xmath39 mev ) ,
the corresponding frequency is @xmath75 the @xmath76 represent the zenith - angle distributions . in our study
we assume two different models . at first
, we consider the `` half - isotropic '' case , where in analogy with a black - body emission , it is assumed @xmath77 for all the species .
we will then compare the results obtained in this widely used prescription , with the one obtained taking the @xmath76 directly from the output of the sn simulations . in this case
, the zenith - angle distributions would be flavor - dependent and forward enhanced ( i.e. peaked at small @xmath31 ) with respect to the half - isotropic emission model ( see fig . 1 and the discussion in @xcite ) .
we will see that the presence of forward peaked distributions will enhance the matter suppression of the maa instability . finally , we comment that our results are based on axial symmetric neutrino angular distributions . in this regard ,
it is interesting to mention that in @xcite it has been shown that if perfect cylindrical symmetry in the initial neutrino distributions were given up , super - fast flavor turnovers could be produced .
we leave the investigation of this interesting issue for a future work .
progenitor mass .
radial evolution of the ratio @xmath78 between the matter @xmath62 and neutrino @xmath61 potentials at different post - bounce times.,scaledwidth=50.0% ] we start our investigation with the case of the @xmath1 @xmath2 iron - core supernova . for this model , the net electron density @xmath12 and the neutrino densities @xmath13 for different post - bounce times were shown in fig .
5 of ref .
@xcite . in order to quantify the relative strength of the matter potential @xmath62 [ eq . ( [ eq : lambda ] ) ] with respect to the neutrino potential @xmath61 [ eq . ( [ eq : mu ] ) ] we plot in fig .
1 the ratio @xmath79 as a function of the radial coordinate @xmath14 at different post - bounce times @xmath80 in the range [ 0.1,0.6 ] s. we realize that @xmath81 before the abrupt discontinuity associated with the shock front position . as we will see with the stability analysis , this strong matter dominance would prevent the flavor conversions before the shock front .
however , for some time snapshots ( i.e. @xmath82 s ) the ratio can become @xmath83 after the shock front , leaving the possibility of flavor conversions in this region .
for the same post - bounce times of fig . 1 , we show in fig .
2 the radial evolution of the eigenvalue @xmath84 determined from the solution of eq .
( [ consit ] ) .
we consider the following cases : _ ( a ) _ @xmath85 and a half - isotropic neutrino emission ( dashed curves ) , _ ( b ) _ dense matter effects and a half - isotropic neutrino emission ( continuous curves ) and , _ ( c ) _ dense matter effects and non - trivial neutrino angular distributions ( dotted curves ) . in fig .
3 we show the survival probability @xmath86 for electron antineutrinos @xmath87 for the same cases of fig . 2 , obtained solving the non - linear propagation equations [ eq .
( [ eq : eom2 ] ) ] .
we start discussing our results for the case of @xmath85 [ case _ ( a ) _ ] .
we realize that when the neutrino system enters the unstable regime ( @xmath67 ) , the @xmath84 function rapidly grows from zero to a peak value greater than one . only for @xmath88 s @xmath89 at the peak .
indeed , for this time the asymmetry parameter @xmath90 ( see fig . 3 in @xcite ) .
then the consistency condition eq .
( [ consit ] ) in order to be satisfied requires a smaller @xmath84 . comparing the results of fig . 2 and 3 one finds a good agreement between the numerical onset of the self - induced flavor conversions triggered by the maa effect and the position of the peak in the @xmath84 function .
we now discuss the situation of realistic matter density profiles and a half - isotropic neutrino emission [ case _ ( b ) _ ] .
as expected , the flavor instability is strongly suppressed with respect to the previous case without matter . in particular , the @xmath84 function , when not completely vanishing ( as at @xmath91 s ) , would start growing only after the shock front position [ see fig . 1 ] .
this is due to the fact that at lower radii the ratio @xmath92 .
the @xmath84 function then reaches peak values between 0.5 and 1 only at intermediate times , i.e. @xmath93 s , for which the ratio @xmath94 in the post - shock region . for the other time snapshots @xmath84 it
is more suppressed , consistently with a larger value of @xmath78 .
comparing these results with the numerical calculation of the @xmath86 in fig . 3 , we realize that the presence of a non - zero @xmath84 is not enough to guarantee the onset of flavor conversions .
indeed , for @xmath95 s , the @xmath84 function is too small and dies out too quickly before triggering flavor conversions . for the cases in which flavor conversions occur , i.e. at @xmath96 s ,
the numerical onset is shifted at larger radii by few hundred km , with respect to the peak of the @xmath84 function .
this delay is due to the fact that since the instability is weaker with respect to the case with @xmath85 , the slower rate of growth implies a larger radial distance in order to develop significant effects on the @xmath86 .
we checked that a non - zero @xmath84 corresponds to the exponential growth of the off - diagonal components @xmath97 of the flux matrices [ see eq .
( [ eq : phi ] ) ] , while the change of the diagonal components , would occur only at larger radii
. it would be interesting to see if the stability analysis can be further developed in order to achieve a better understanding of this dynamics .
then we consider the case in which also the flavor - dependent forward - peaked neutrino angular distributions are taken into account [ case _ ( c ) _ ] .
we find that the @xmath84 function in this case is completely suppressed .
this is consistent with the expectation that the @xmath6-@xmath6 strength is weaker for forward - peaked distributions , making more effective the matter suppression .
this result is consistent with the output of the numerical simulations that show for all the considered time snapshots @xmath98 .
finally we mention that in @xcite it has been claimed that possible residual scatterings could affect @xmath6 s after the neutrinosphere , producing a small `` neutrino halo '' that would broaden the @xmath6 angular distributions @xcite at @xmath99 km .
we checked that the results of the sn simulations we are using have not enough angular resolution to exhibit this feature .
however , in order to characterize the possible halo effect , we performed the same analytical estimation presented in @xcite .
we repeated the stability analysis including the halo effect in the @xmath6 angular distributions , without finding any change with respect to the results shown here .
therefore , we conclude that for our @xmath1 @xmath2 sn model , maa instability is always suppressed by the dense matter effects during the accretion phase .
progenitor mass .
radial evolution of the @xmath84 function ( in units of km@xmath63 ) at different post - bounce times with @xmath85 for a half - isotropic neutrino emission ( dashed curves ) and in presence of matter effects , with a half - isotropic neutrino emission ( continuous curves ) and with flavor - dependent angular distributions ( dotted curves).,scaledwidth=50.0% ] progenitor mass .
radial evolution of the survival probability @xmath86 for electron antineutrinos at different post - bounce times for the maa evolution with @xmath85 for a half - isotropic neutrino emission ( dashed curves ) and in presence of matter effects , with a half - isotropic neutrino emission ( continuous curves ) and with flavor - dependent angular distributions ( dotted curves).,scaledwidth=50.0% ] progenitor mass .
radial evolution of the ratio @xmath78 between the matter @xmath62 and neutrino @xmath61 potentials at different post - bounce times.,scaledwidth=50.0% ] in this section we analyze the flavor conversions for the model of @xmath3 @xmath2 o - ne - mg progenitor . in fig . 4 we plot the ratio @xmath79 for the same time snapshots of fig .
9 of @xcite , i.e. in the range [ 0.08,0.25 ] s. we realize that in this case there is no abrupt discontinuity associated with the shock front . indeed , for this low - mass progenitor there is no extended accretion phase , since the explosion succeeds very shortly after the core - bounce .
therefore , the shock - front is already beyond the radial range interesting for the flavor conversions . since in this case
the matter density of the envelope is low compared to the iron - core progenitors , the electron density profile above the core is very steep .
therefore , at @xmath100 few hundred km , the ratio @xmath101 [ for @xmath102~s$ ] it monotonically decreases becoming also smaller than 1 ] suggests that flavor conversions could arise there . in fig . 5 we show the radial evolution of the eigenvalue @xmath84 for the time snapshots shown in fig . 4 , determined from the solution of eq .
( [ consit ] ) .
we use the same format of fig . 2 . in fig .
6 we show the corresponding survival probability @xmath86 for electron antineutrinos @xmath87 . starting with the case without matter term , i.e. @xmath85 ( dashed curves ) , we see that the @xmath84 function rapidly becomes larger than 1 , and the peak corresponds to the onset of the flavor conversions in fig .
we pass now considering the case with @xmath62 and half - isotropic neutrino angular distributions ( continuous curves ) .
we realize that , for @xmath103 s in the region where @xmath84 grows , @xmath104 . therefore , the matter suppression of the instability is never complete .
moreover , the rise of @xmath84 is rapid and the position of the peak corresponds to the onset of the flavor conversions seen in fig .
conversely , for the other time snapshots ( @xmath105 s ) where @xmath106 , the suppression of the instability is stronger . moreover , the @xmath84 curves are broadened and there is not a clear peak .
therefore , one can not easily link a non - zero @xmath84 with the numerical onset of the flavor conversions . in the case of @xmath62 and flavor
dependent forward - peaked neutrino angular distributions ( dotted curves ) , as expected we find a stronger suppression in the flavor instability .
however , as shown in fig .
1 of @xcite , the angular spectra of different flavors for the 8.8 m@xmath107 sn are significantly less forward - peaked than in the case of the 10.8 m@xmath107 sn .
therefore , their effect is less pronounced in this case . in particular , for @xmath108 s , @xmath84 is large enough to trigger flavor conversions .
conversely , these are strongly inhibited at @xmath109 s , and completely suppressed for @xmath110 s. finally , we checked also in this case that including a possible halo effect does not change the results of the stability analysis . in conclusion , for our sn model with 8.8 m@xmath107 progenitor mass the matter suppression of the maa instability is not complete at early times . therefore , in principle one would expect interesting time - dependent features in the observable neutrino spectra .
progenitor mass .
radial evolution of the @xmath84 function ( in units of km@xmath63 ) at different post - bounce times with @xmath85 for a half - isotropic neutrino emission ( dashed curves ) and in presence of matter effects , with a half - isotropic neutrino emission ( continuous curves ) and with flavor - dependent angular distributions ( dotted curves).,scaledwidth=50.0% ] progenitor mass .
radial evolution of the survival probability @xmath86 for electron antineutrinos at different post - bounce times for the maa evolution with @xmath85 for a half - isotropic neutrino emission ( dashed curves ) and in presence of matter effects , with a half - isotropic neutrino emission ( continuous curves ) and with flavor - dependent angular distributions ( dotted curves).,scaledwidth=50.0% ]
we have performed a dedicated study of the matter suppression of the maa instability , connected with the axial symmetry breaking in the self - induced oscillations , during the accretion phase for two sn models with different progenitor masses .
we characterize the sn densities and the neutrino angular spectra with results from recent sn hydrodynamical simulations .
we compared the linear stability analysis with the numerical results of the flavor evolution in which we have looked at a local solution of the equations of motions along a specific line of sight .
for the case of an iron - core 10.8 m@xmath107 we found that during the accretion phase the dominant matter density strongly suppresses the maa instability .
in particular , including realistic forward - peaked @xmath6 angular distributions significantly reduces the strength of the @xmath6-@xmath6 interaction term . as a result , in this case the matter suppression of the self - induced flavor conversions would be complete . in the case of a low - mass o - ne - mg sn with 8.8 m@xmath107 progenitor , where the accretion phase is extremely short ,
the matter density profile is lower and the @xmath6 angular distributions less forward - peaked than in iron - core models . as a consequence
, we found that also with realistic angular distributions flavor conversions would be possible at early times , producing in principle interesting time - dependent modulations .
our analysis is complementary to previous studies @xcite , where some of us explored the matter suppression of self - induced flavor conversions in inverted neutrino mass hierarchy , in axial - symmetric models ( see also @xcite ) .
the complete suppression of the self - induced effects in both the mass hierarchies for iron - core sne , implies that the neutrino signal during the accretion phase will be processed only by the ordinary mikheyev - smirnov - wolfenstein effect in the outer stellar layers .
this effect would allow in principle to distinguish the neutrino mass hierarchy through the study of the rise time of the sn neutrino signal @xcite .
the phenomenological importance of our findings motivates further studies with other sn models to confirm the generality of our results .
in particular , an accurate characterization of the neutrino angular distributions seems necessary in order to get accurate predictions on the matter suppression . at this regard , in
@xcite it has been shown that if perfect cylindrical symmetry in the initial neutrino distributions were given up , then super - fast flavor turnovers could occur .
these effects can not be tested within our spherically symmetric sn model .
however , recently three - dimensional sn simulations have been carried on , characterizing the neutrino signal during the accretion phase .
surprisingly , a lepton - emission asymmetry among different flavor has been found @xcite .
in particular , the electron ( anti)neutrino fluxes show a dipole structure , while the @xmath111 are almost spherically symmetric .
we plan to investigate in a future work the role of the matter suppression in this flavor configuration , including also the not axisymmetric neutrino and matter angular distributions .
self - induced flavor conversions associated with the maa instability would still be possible for o - ne - mg sne during the accretion phase , and possibly for iron core sne during the cooling phase , when the matter term becomes sub - dominant with respect to the neutrino - neutrino interaction term . in these situations , a self - consistent treatment of the neutrino equations of motion considering also the flavor evolution along
the transverse direction is still lacking .
this would imply passing from an ordinary to a partial differential equation problem , adding a big layer of complication in the solution of the equations of motion .
this effort is well motivated by the perspective of getting an accurate characterization of the sn neutrino spectral features that would be observable in the planned large underground neutrino detectors @xcite .
we thank t. fischer , g. raffelt , s. sarikas and m. wu for useful discussions .
s.c . acknowledges support from the european union through a marie curie fellowship , grant no .
piif - ga-2011 - 299861 , and through the itn `` invisibles '' , grant no .
pitn - ga-2011 - 289442 .
the work of a.m. was supported by the german science foundation ( dfg ) within the collaborative research center 676 `` particles , strings and the early universe . ''
n.s . acknowledges support from the european union fp7 itn invisibles ( marie curie actions , pitn- ga-2011- 289442 ) .
d.s . acknowledges support by the fundao para a cincia e tecnologia ( portugal ) under grant sfrh / bd/66264/2009 .
h. duan , g. m. fuller , j. carlson and y. z. qian , `` simulation of coherent non - linear neutrino flavor transformation in the supernova environment .
i : correlated neutrino trajectories , '' phys .
d * 74 * , 105014 ( 2006 ) [ astro - ph/0606616 ] .
s. hannestad , g. g. raffelt , g. sigl and y. y. y. wong , `` self - induced conversion in dense neutrino gases : pendulum in flavour space , '' phys .
d * 74 * , 105010 ( 2006 ) [ erratum - ibid .
d * 76 * , 029901 ( 2007 ) ] [ astro - ph/0608695 ] .
g. l. fogli , e. lisi , a. marrone and a. mirizzi , `` collective neutrino flavor transitions in supernovae and the role of trajectory averaging , '' jcap * 0712 * , 010 ( 2007 ) [ arxiv:0707.1998 [ hep - ph ] ] . g. l. fogli , e. lisi , a. marrone , a. mirizzi and i. tamborra , `` low - energy spectral features of supernova ( anti)neutrinos in inverted hierarchy , '' phys .
d * 78 * , 097301 ( 2008 ) [ arxiv:0808.0807 [ hep - ph ] ] .
h. duan , g. m. fuller and y. -z .
qian , `` collective neutrino oscillations , '' ann .
nucl . part .
sci . * 60 * , 569 ( 2010 ) [ arxiv:1001.2799 [ hep - ph ] ] .
a. esteban - pretel , s. pastor , r. toms , g. g. raffelt and g. sigl , `` decoherence in supernova neutrino transformations suppressed by deleptonization , '' phys . rev .
d * 76 * , 125018 ( 2007 ) [ arxiv:0706.2498 [ astro - ph ] ] .
a. mirizzi and r. toms , `` multi - angle effects in self - induced oscillations for different supernova neutrino fluxes , '' phys .
d * 84 * , 033013 ( 2011 ) [ arxiv:1012.1339 [ hep - ph ] ] .
a. banerjee , a. dighe and g. raffelt , `` linearized flavor - stability analysis of dense neutrino streams , '' phys .
d * 84 * , 053013 ( 2011 ) [ arxiv:1107.2308 [ hep - ph ] ] .
g. raffelt , s. sarikas and d. d. s. seixas , `` axial symmetry breaking in self - induced flavor conversion of supernova neutrino fluxes , '' phys .
lett . * 111 * , 091101 ( 2013 ) [ arxiv:1305.7140 [ hep - ph ] ] .
s. samuel , `` bimodal coherence in dense selfinteracting neutrino gases , '' phys .
d * 53 * , 5382 ( 1996 ) [ hep - ph/9604341 ] .
g. raffelt and d. d. s. seixas , `` neutrino flavor pendulum in both mass hierarchies , '' phys .
d * 88 * , 045031 ( 2013 ) [ arxiv:1307.7625 [ hep - ph ] ] .
h. duan , `` flavor oscillation modes in dense neutrino media , '' phys .
d * 88 * , 125008 ( 2013 ) [ arxiv:1309.7377 [ hep - ph ] ] .
a. mirizzi , `` multi - azimuthal - angle effects in self - induced supernova neutrino flavor conversions without axial symmetry , '' phys . rev .
d 88 , * 073004 * ( 2013 ) [ arxiv:1308.1402 [ hep - ph ] ] .
a. mirizzi , `` self - induced spectral splits with multi - azimuthal - angle effects for different supernova neutrino fluxes , '' arxiv:1308.5255 [ hep - ph ] .
a. esteban - pretel , a. mirizzi , s. pastor , r. tomas , g. g. raffelt , p. d. serpico and g. sigl , `` role of dense matter in collective supernova neutrino transformations , '' phys .
d * 78 * , 085012 ( 2008 ) [ arxiv:0807.0659 [ astro - ph ] ] .
s. chakraborty , t. fischer , a. mirizzi , n. saviano and r. tomas , `` no collective neutrino flavor conversions during the supernova accretion phase , '' phys .
* 107 * , 151101 ( 2011 ) [ arxiv:1104.4031 [ hep - ph ] ] .
s. chakraborty , t. fischer , a. mirizzi , n. saviano and r. tomas , `` analysis of matter suppression in collective neutrino oscillations during the supernova accretion phase , '' phys .
d * 84 * , 025002 ( 2011 ) [ arxiv:1105.1130 [ hep - ph ] ] .
n. saviano , s. chakraborty , t. fischer and a. mirizzi , `` stability analysis of collective neutrino oscillations in the supernova accretion phase with realistic energy and angle distributions , '' phys . rev .
d * 85 * , 113002 ( 2012 ) [ arxiv:1203.1484 [ hep - ph ] ] .
s. sarikas , g. g. raffelt , l. hudepohl and h. -t .
janka , `` suppression of self - induced flavor conversion in the supernova accretion phase , '' phys .
* 108 * , 061101 ( 2012 ) [ arxiv:1109.3601 [ astro-ph.sr ] ] .
g. sigl and g. raffelt , `` general kinetic description of relativistic mixed neutrinos , '' nucl .
b * 406 * , 423 ( 1993 ) .
l. wolfenstein , `` neutrino oscillations in matter , '' phys .
d * 17 * , 2369 ( 1978 ) ; s. p. mikheev and a. yu .
smirnov , `` resonance enhancement of oscillations in matter and solar neutrino spectroscopy , '' yad .
* 42 * , 1441 ( 1985 ) [ sov .
j. nucl .
42 * , 913 ( 1985 ) ] .
t. fischer , s. c. whitehouse , a. mezzacappa , f. -k .
thielemann and m. liebendorfer , `` protoneutron star evolution and the neutrino driven wind in general relativistic neutrino radiation hydrodynamics simulations , '' astron .
astrophys .
* 517 * ( 2010 ) a80 [ arxiv:0908.1871 [ astro-ph.he ] ] .
j. f. cherry , j. carlson , a. friedland , g. m. fuller and a. vlasenko , `` neutrino scattering and flavor transformation in supernovae , '' phys .
* 108 * , 261104 ( 2012 ) [ arxiv:1203.1607 [ hep - ph ] ] .
s. sarikas , i. tamborra , g. raffelt , l. hudepohl and h. -t .
janka , `` supernova neutrino halo and the suppression of self - induced flavor conversion , '' phys .
d * 85 * , 113007 ( 2012 ) [ arxiv:1204.0971 [ hep - ph ] ] .
d. serpico , s. chakraborty , t. fischer , l. hudepohl , h. -t .
janka and a. mirizzi , `` probing the neutrino mass hierarchy with the rise time of a supernova burst , '' phys .
d * 85 * , 085031 ( 2012 ) [ arxiv:1111.4483 [ astro-ph.sr ] ] .
i. tamborra , f. hanke , h. -t .
janka , b. mueller , g. g. raffelt and a. marek , `` self - sustained asymmetry of lepton - number emission : a new phenomenon during the supernova shock - accretion phase in three dimensions , '' arxiv:1402.5418 [ astro-ph.sr ] . | it has been recently pointed out that removing the axial symmetry in the `` multi - angle effects '' associated with the neutrino - neutrino interactions for supernova ( sn ) neutrinos , a new multi - azimuthal - angle ( maa ) instability would arise . in particular , for a flux ordering @xmath0 , as expected during the sn accretion phase , this instability occurs in the normal neutrino mass hierarchy .
however , during this phase the ordinary matter density can be larger than the neutrino one , suppressing the self - induced conversions . at this regard
, we investigate the matter suppression of the maa effects , performing a linearized stability analysis of the neutrino equations of motion , in the presence of realistic sn density profiles .
we compare these results with the numerical solution of the sn neutrino non - linear evolution equations .
assuming axially symmetric distributions of neutrino momenta we find that the large matter term strongly inhibits the maa effects .
in particular , the hindrance becomes stronger including realistic forward - peaked neutrino angular distributions . as a result , in our model for a @xmath1 @xmath2 iron - core sne
, maa instability does not trigger any flavor conversion during the accretion phase . instead ,
for a @xmath3 @xmath2 o - ne - mg core sn model , with lower matter density profile and less forward - peaked angular distributions , flavor conversions are possible also at early times . |
the nonlinear coupled set of three odes [ lorenz ] x = -x + y , y = r x - y -xz , z = -bz + xy , was first derived by berry saltzman and edward lorenz @xcite as a model for thermal convection between two plates perpendicular to the direction of the earth s gravitational force .
it has three control parameters .
first , there is the normalized temperature difference @xmath5 between the hotter lower plate and the colder upper one .
@xmath6 is the temperature difference at the onset of convection .
@xmath7 , the rayleigh number , is the dimensionless temperature difference .
@xmath8 denotes the gravitational acceleration , @xmath9 the isobaric thermal expansion coefficient , @xmath10 the distance between the plates in upward direction , and @xmath11 , @xmath12 characterize the kinematic viscosity and thermal diffusivity , respectively .
second , there is the prandtl number @xmath13 , describing the ratio of the two molecular time scales @xmath14 and @xmath15 for momentum and energy transport in the system . @xmath1
also measures the ratio of the widths @xmath16 and @xmath17 of the kinematic and the thermal boundary layers .
large prandtl numbers @xmath1 indicate a broad kinematic boundary layer containing a small thermal one , fast molecular momentum exchange and small convective reynolds numbers .
small prandtl numbers characterize fast molecular thermal exchange , a broad thermal boundary layer containing a thinner kinematic one and larger convective reynolds numbers .
typical values for @xmath1 are @xmath18 in air ( at room temperature ) , n@xmath19 , o@xmath19 , he , and other gases , about 7 in water ( room temperature ) , @xmath20 through @xmath21 in typical organic liquids as e.g. dipropylene glycol ( @xmath20 ) , or @xmath22 in glycerine .
small prandtl numbers are realized in mercury ( @xmath23 ) or liquid sodium with @xmath24 .
the third parameter in eqs .
( [ lorenz ] ) is the coupling strength @xmath25 , originating from the nonlinearity @xmath26 of the boussinesq equation .
it is of order @xmath27 and conventionally chosen as @xmath28 ; we adopt this value here .
the variables @xmath29 , and @xmath30 describe , respectively , the amplitude of the velocity mode ( the only one left in the model ) , the amplitude of the corresponding temperature mode , and the amplitude of the mode measuring the nondimensional heat flux @xmath31 , the nusselt number . @xmath32 characterize the roll pattern , the heat flux is nu = 1 + 2 z / r .
[ nusselt ] for @xmath33 there exists one stable fixed point p@xmath34 ( @xmath35 ) , with @xmath36 , corresponding to pure molecular heat conduction . for @xmath0
exceeding @xmath27 , convective rolls develop , described by two stable fixed points [ fixedpoints ] _
= ( x_,y_,z _ ) = ( , , r-1 ) hence @xmath37 . the trajectories in phase space @xmath38 are spirals towards these fixed points . stability is lost for r > r_c = .
[ stabiloss ] for the canonical choice @xmath39 it is @xmath40 . at such large @xmath0
the three mode approximation for the pdes describing thermal convection has of course ceased to be physically realistic , but mathematically the model now starts to show its most fascinating properties , because a strange attractor develops for @xmath41 . with further increasing @xmath0 , windows of regular behavior
emerge out of chaos where stable periodic orbits exist .
these stability windows tend to become broader and broader until for sufficiently large @xmath0 chaos seems to disappear and one type of stable periodic solution takes over . of the numerous studies of the lorenz model we cite as examples the monograph @xcite and the more recent survey in jackson s book @xcite . to the best of our knowledge , the parameter dependence of the lorenz model has mainly been analyzed with respect to its control parameter @xmath0 , the external forcing , and sometimes with varying @xmath25 ( as in the work @xcite which is cited in @xcite ) .
the prandtl number @xmath1 has usually been fixed to its canonical value @xmath39 , and as a rule , @xmath25 has been taken as @xmath42 .
the limit of large @xmath0 at arbitrary fixed @xmath1 was thoroughly studied by robbins @xcite and sparrow @xcite . in the present paper
we study the parameter range where both @xmath0 _ and _ @xmath1 become large .
the scheme of bifurcations is then a very different one . with @xmath1 as large as 60 ( fig .
[ fig : r409sigma60 ] ) one is just beginning to see a structure that unfolds on much larger scales ( fig .
[ fig : rsigma2000 ] ) .
there is ongoing repetition of a few basic features , but these are different from the structure at low @xmath1 .
the line @xmath43 divides these two regimes as we shall show .
our theoretical explanation of these findings focuses on a backbone structure of the dominant stable periodic attractors , but orbits with arbitrarily high periods , and chaos , are also observed as part of the pattern .
an important feature of the lorenz equations is their invariance with respect to the reflection : x , y -x , -y z z .
[ symmetry1 ] evidently the @xmath44-axis is a @xmath45-invariant line .
for @xmath46 it is also the stable manifold of the fixed point p@xmath34 .
this characteristic symmetry property with respect to @xmath45 has implications on the nature of the period doubling scenarios which occur over and over again as the parameters @xmath0 , @xmath1 are varied : they do _ not _ follow the familiar scheme of the paradigmatic logistic iteration @xmath47 but rather the somewhat more intricate scheme of the cubic iteration @xmath48 .
this cubic mapping possesses the symmetry @xmath49 , with @xmath50 , and may be viewed as one building block for understanding the transition to chaos in the lorenz system , see hao et al .
@xcite for a detailed analysis of this aspect .
the motivation of the present work was the simple question why the meteorologist lorenz would have chosen the prandtl number @xmath39 , which is adequate for water but less so for air with its order of magnitude @xmath51 .
the reader will easily find that taking @xmath1 this small , he will no longer find chaotic solutions .
this is confirmed by a glance on eq .
( [ stabiloss ] ) : with @xmath52 , the critical parameter @xmath53 lies in the unphysical range @xmath54 . if one chooses , instead , some @xmath55 , implying @xmath56 , the window structure in the @xmath0-dependence is similar to the case @xmath39 although the details may be nontrivially different .
but one hardly sees how these differences in the @xmath0-behavior for some discrete @xmath1-values are connected .
this motivated us to vary @xmath1 in sufficiently small steps and study the 2-dimensional @xmath57-space of control parameters in a larger domain than has hitherto been done .
however , quite apart from such a physical motivation , the mathematical complexity of this innocent looking set of equations warrants an investigation of their asymptotic behavior where it has not yet been explored . in section [ sec : two ] we present our parameter survey ( phase diagrams ) and discuss its main features .
then , in section [ sec : attractors ] the nature of two kinds of periodic orbits is described whose stability regions define the dominant regular features of the large scale phase diagram ; an analytic derivation of these observations is given in section [ secparameterspace ] .
more complex aspects of the lorenz dynamics such as bifurcations of periodic orbits , chaos , and coexistence of attractors , are discussed in section [ secbifs ] .
we point out regularities , and similarities to other well known systems , but do not give detailed explanations of these features .
a glance at figs . [ fig : r409sigma60 ] and [ fig : rsigma2000 ] gives an impression of the bifurcation scheme on large and very large scales of @xmath0 ( abscissa ) and @xmath1 .
the parameter @xmath25 , as mentioned already , is kept at its standard value @xmath42 .
colors indicate the existence of attractors of a certain type , which does not preclude the coexistence of others .
more precisely , the figures are parameter scans where each point @xmath2 is colored according to which type of attractor is approached from a fixed initial point @xmath58 .
such pictures are called final state diagrams , or phase diagrams .
the simplest attractors are the fixed points p@xmath59 , see ( [ fixedpoints ] ) .
their stability is lost along the curve @xmath60 , see ( [ stabiloss ] ) , or rather by its inverse function @xmath61 with [ hyperbola ] _
c(r ) = - . for large @xmath0 ,
the upper and lower asymptotes @xmath62 and @xmath63 are , respectively , [ sigmaul ] _ u = r - 2(b+2 ) _ = b + 1 .
the next order correction is @xmath64 , describing the approach of the upper asymptote from below and the lower one from above .
the @xmath2-region where the points p@xmath59 are stable according to this linear analysis is colored black , the rest is left white , before the other colors are added in figs . [
fig : r409sigma60 ] and [ fig : rsigma2000 ] .
these colors code for the periods of attractors which are determined by the procedure described below : yellow for orbits of period 1 , red for period 2 , green for period 4 , blue for period 8 , grey for period 3 , olive for period 6 .
@xmath2-phase diagram in the range @xmath65 and @xmath66 .
colors indicate the type of attractor reached from the initial condition @xmath67 : fixed points p@xmath59 ( black ) , orbits of period 2 ( red ) , period 4 ( green ) , period 8 ( blue ) , chaotic attractors ( white ) .
this picture is a zoom into the lower left corner of fig .
[ fig : rsigma2000 ] . the line @xmath68 in color magenta marks a transition range between different global structures for smaller or larger @xmath1 .
further details are discussed in the text , section [ ssec : types ] .
, scaledwidth=50.0% ] the parameter ranges are @xmath69 .
the color code is the same as in fig .
[ fig : r409sigma60 ] .
yellow bands indicate stability of orbits with period 1 ( not present in the zoomed lower left corner , cf .
[ fig : r409sigma60 ] ) .
the dots in the red and yellow bands along the line @xmath70 are , from bottom left to top right , s@xmath71 , a@xmath71 , s@xmath19 , a@xmath19 , , defined according to for @xmath72 . with their upper parts ,
the bands penetrate well into the range above @xmath73 , cf .
( the line can be identified as the border to the black region through the bands ) , indicating coexistence of periodic orbits with the stable fixed points p@xmath74 . ]
( color online ) .
( a1 ) : @xmath75-projection of an asymmetric attractor of period 3 ( abscissa @xmath76 , ordinate @xmath44 ) .
@xmath77 , @xmath78 , initial conditions @xmath79 , @xmath80 .
the fat points are the fixed points p@xmath59 .
( a2 ) : the same for a chaotic attractor ; @xmath81 , @xmath82 .
( b1 ) and ( b2 ) : corresponding poincar sections with the @xmath83-plane at @xmath84 ; intersection points are grey ( green ) if from below , black ( red ) if from above ( abscissa @xmath85 , ordinate @xmath76).,scaledwidth=60.0% ] for each @xmath2 on a grid of @xmath86 points , an orbit was started with initial conditions @xmath67 .
a transient regime of 80000 iterations was allowed to elapse before it was assumed that an attractor was reached .
the trajectory was then analyzed in terms of its intersections with the @xmath83-plane at @xmath87 which is an appropriate poincar surface of section .
the period p of a periodic orbit is defined as half the number of intersection points of the orbit with that plane ( in one half the trajectory intersects from above , in the other half from below ) .
[ fig : poinc ] gives two examples for illustration . to determine the period we used a rather simple scheme .
a disk of radius @xmath88 was placed around each point of intersection unless it fell in a disk already generated .
when the number of patches so obtained remained finite , like in the left part of fig .
[ fig : poinc ] , the orbit was assumed to be periodic .
when , on the other hand , the number of disks seemed to grow indefinitely , like in the right part of the figure , the orbit was considered to be chaotic . in practice
we waited for more than 100 intersections , but only periods up to 8 form clearly identifiable bands on the scales of the figures .
when a periodic orbit was detected , the color of the grid point @xmath2 was overwritten with the color of the code given above . otherwise the original black or white color was left unchanged .
hence chaotic attractors appear as black or white points .
the method produces two artifacts .
first , it can not identify the fixed point attractors p@xmath59 ; hence , in the black region it does not distinguish between fixed points and chaotic attractors .
second , when the relaxation towards an attractor is very slow , the 80000 transient iterations may not be sufficient to reach it , and the method produces more than 100 patches .
hence this situation is not distinguished from chaos .
it is observed in connection with critical slowing down in the neighborhood of a bifurcation of the attractor .
examples are the white bands in the red regions at bottom and top of fig .
[ fig : r409sigma60 ] , the narrow seams between red and green regions , and the lines inside the two green bands in the chaotic range . the nature of the corresponding bifurcations will be elucidated in the following section .
since only one initial condition is used , the procedure does not determine whether different initial conditions would lead to different attractors for the chosen parameters @xmath2 .
the question of coexistence of attractors will be addressed later , in sec .
[ ssec : coexistence ] . at this point
we can assert already that where a colored band extends into the black region , like in the lower right and upper left parts of fig .
[ fig : r409sigma60 ] , the corresponding periodic orbit coexists with the fixed points p@xmath59 . the arrangement of bands in fig .
[ fig : r409sigma60 ] exhibits a qualitative structural symmetry with respect to a line which has been added in magenta color .
this line , mathematically defined as @xmath43 , separates a large-@xmath1 structure from a kind of mirror image in the low-@xmath1 part of the phase diagram .
it is conspicuous how every band below @xmath89 seems to have a counterpart on the other , upper side .
but , as a glance on fig .
[ fig : rsigma2000 ] shows , this `` symmetry '' disappears on a larger scale : there is no repetitive structure in the low-@xmath1 region , in contrast to the large-@xmath1 range .
no orbits of period 1 ( yellow bands ) are observed in the @xmath2-range covered in fig .
[ fig : r409sigma60 ] .
however , in fig .
[ fig : rsigma2000 ] which covers a much larger @xmath57-range , an onion like repetitive pattern of yellow ( period 1 ) and red ( period 2 ) bands dominates in the upper part .
these bands are separated from each other by chaotic regions which also contain small bands of higher periods .
the bands ( above @xmath90 ) have an hyperbolic shape whose upper branches seem to have asymptotic slopes of approximately 1.3 while the lower branches will be shown in section [ secparameterspace ] to have a limiting behavior @xmath91 like the `` bottom line '' @xmath89 .
the upper branches penetrate significantly into the region @xmath92 where the fixed points p@xmath59 are stable , see eq .
( [ sigmaul ] ) .
this means there is coexistence of fixed point attractors and periodic attractors . to generate fig .
[ fig : rsigma2000 ] the initial conditions were @xmath93 and @xmath94 . choosing the initial conditions as in fig .
[ fig : r409sigma60 ] , the corresponding picture ( not shown here ) looks different in that the greater part of the white bands of chaos is replaced by colors yellow or red .
this reflects the coexistence of chaotic and periodic attractors .
the bands have infinite vertical slopes along the line @xmath95 . on this line ,
we have marked points s@xmath71 , a@xmath71 , s@xmath19 , a@xmath19 , @xmath96 , ... in the red ( period 2 ) and yellow ( period 1 ) bands , respectively .
the names of these points will also be taken to denote the corresponding bands because the attractors of a given band are topologically identical .
their nature will first be described phenomenologically in the subsequent section , where we shall also see how it changes from one band to the next .
an analytic explanation of the findings will then be given in section [ secparameterspace ] .
in this section we explore the nature of the dominant periodic attractors , i.e. , of the orbits with periods 1 or 2 , by means of numerical and graphical investigation . their bifurcations into orbits of higher periods , as well as into strange attractors , will be discussed in section [ secbifs ] . to begin with ,
consider the attractors corresponding to the @xmath2-values of a@xmath97 in fig .
[ fig : attractora3 ] and of s@xmath98 in fig .
[ fig : attractors4 ] , in the three standard projections .
comparison with the s@xmath71 , a@xmath71 , s@xmath19 , and a@xmath19 attractors ( see figs . [
fig : winding ] and [ fig : rxbifdia ] ) shows that on the scales used , @xmath99 , all a@xmath100 attractors look alike , and so do the orbits of s@xmath100 .
furthermore , the attractor of the parameter point s@xmath100 ( for given @xmath101 ) looks like a combination of the asymmetric orbits of a@xmath100 and its @xmath45-mirror image @xmath102 .
the time course of the variables @xmath85 , @xmath76 , and @xmath44 is plotted in fig .
[ fig : a3s4intime ] , both for the a@xmath97 attractor ( top ) and the s@xmath98 one ( bottom ) .
the combined picture shows the typical behavior of a relaxation oscillator .
two different regimes can clearly be distinguished : first , the variable @xmath44 decreases slowly from an upper value @xmath103 to a lower value @xmath104 , with @xmath105 , while @xmath85 and @xmath76 are virtually zero .
then , in a very fast process , @xmath85 and @xmath76 start out on a large excursion with amplitude @xmath106 , while @xmath44 jumps back to its upper value .
the @xmath75-projection exhibits an almost perfect half circle ; the @xmath83-projection shows that during the excursion , @xmath85 lags somewhat behind @xmath76 .
stable asymmetric periodic orbit corresponding to the parameters @xmath107 of the point @xmath108 in fig .
[ fig : rsigma2000 ] . from left to right , the attractor is shown in @xmath83 , @xmath109 , and @xmath75 projection ( the abscissa is always mentioned as first coordinate , the ordinate as second ) .
the dots mark the projections of the fixed points @xmath110 , see ( [ fixedpoints ] ) .
the arrows indicate the direction of flow ; the decrease of @xmath44 takes place very close to the @xmath44-axis .
the attractor coexists with its @xmath45-partner which is obtained by reflection of @xmath85 and @xmath76 .
together these two attractors look like the symmetric attractor in fig .
[ fig : attractors4 ] .
the scales are @xmath111.,scaledwidth=75.0% ] stable symmetric periodic orbit corresponding to @xmath112 , the point @xmath113 in fig .
[ fig : rsigma2000 ] .
same projections as in fig .
[ fig : attractora3 ] .
the scales are @xmath114.,scaledwidth=75.0% ] ( color online ) .
time development of @xmath85 , @xmath76 , and @xmath44 within one unit of time @xmath115 ( abscissa ) , for the attractors of @xmath108 ( top ) and of @xmath113 ( bottom ) .
the upper parts show the @xmath44-coordinates , the lower parts @xmath85 and @xmath76 which are nearly indistinguishable on this time scale .
the vertical scale is @xmath116.,scaledwidth=80.0% ] ( color online ) .
@xmath83-projections of ( from left to right ) the attractors of s@xmath71 , a@xmath71 , s@xmath19 . in the case of a@xmath71
the two asymmetric orbits are shown in different color .
the insert in the left picture is a blow - up of the neighborhood of the origin .
the @xmath2 values are s@xmath71 = ( 172,130 ) , a@xmath71 = ( 402,302 ) , s@xmath19 = ( 632,474 ) ; the scales are @xmath117 . , scaledwidth=75.0% ] ( color online ) .
successive blow - ups of the box in the middle picture of fig .
[ fig : winding ] .
the sequence demonstrates that the two orbits of a@xmath71 do not reach the origin ; rather each of them winds once around the @xmath44-axis .
, scaledwidth=75.0% ] from figs .
[ fig : attractora3]-[fig : a3s4intime ] it is not obvious what happens during the slow phase where the orbits move down along the @xmath44-axis .
this question is clarified by the sequence of @xmath83-projections in figs .
[ fig : winding ] and [ fig : a1blowup ] . the individual pictures in fig .
[ fig : winding ] show the attractors of s@xmath71 , a@xmath71 , s@xmath19 ; the other projections would be similar to those in figs .
[ fig : attractora3 ] and [ fig : attractors4 ] . taken together ,
the pictures make it clear that the orbits wind around the @xmath44-axis on spirals with decreasing @xmath118 . coming from large values of @xmath119 , near the maximum value of @xmath44
, they enter a close neighborhood of the @xmath44-axis when they pass a point where @xmath120 . after a quarter turn in clockwise direction ( looking down the @xmath44-axis ) there
comes a point where @xmath121 , at smaller values of @xmath44 and @xmath119 .
what happens next depends on the index @xmath101 of s@xmath100 or a@xmath100 . in the case of s@xmath71
the orbit quickly leaves the @xmath83-origin , cf .
the inset in the left part of fig .
[ fig : winding ] ; it performs a large loop and returns to the @xmath83-origin region from the other side , but again near the maximum of @xmath44 .
all in all it encircles the @xmath44-axis once , its winding number @xmath122 with respect to the @xmath44-axis being equal to the number @xmath123 of zeros in @xmath76 ( and correspondingly in @xmath85 ) upon each close encounter .
in the case of a@xmath71 , the orbit stays longer near the @xmath44-axis , with a second zero of @xmath76 and also of @xmath85 , each with smaller values of @xmath119 ( and @xmath44 ) ; this is illustrated by the sequence of zooms in fig .
[ fig : a1blowup ] .
its winding number with respect to the @xmath44-axis is also @xmath124 , but now there are two of these orbits , shown in different colors . to keep track of the winding number @xmath122 for increasing band label @xmath101 , we may focus on the number @xmath125 of zeros of @xmath76 ( or @xmath126 upon close encounter with the @xmath44-axis .
detailed observation shows that this number is [ windingnumbers ] w(_n ) = n(_n ) = 2n-1
w(_n)=n(_n ) = n . fig . [
fig : a3s4logintime ] shows a logarithmic plot of @xmath127 and @xmath128 for a@xmath97 ( @xmath129 ) and s@xmath98 ( @xmath130 ) .
we observe three different regimes of time dependence . soon after @xmath44 reaches its peak value , at a time @xmath131 ,
the first zeros of @xmath76 and @xmath85 occur , visible as logarithmic singularities .
the first regime of oscillatory exponential decay of @xmath85 and @xmath76 ends at a time @xmath132 . in the second phase
there are no further @xmath85- or @xmath76-zeros , but the @xmath44-decay continues with deceleration up to a time @xmath133 where the distance @xmath119 of the orbit from the @xmath44-axis reaches a minimum .
the third regime is one of explosive , i.e. , faster than exponential increase to the original values of @xmath32 and @xmath44 .
the spiraling behavior in the first regime seems reminiscent of `` shilnikov chaos '' @xcite .
however , in this case all the relevant dynamics stays far away from the fixed point .
the invariant line @xmath134 is part of the stable manifold of the origin , but this is not important for our arguments .
the main ingredient is that it is attractive for @xmath135 ( i.e. up to time @xmath133 ) and repulsive for @xmath136 .
long before the orbit gets close to the origin it leaves the @xmath44-axis along its unstable eigenvector , with @xmath137 .
( color online ) .
time development of @xmath138 ( red ) , @xmath139 ( blue ) , and above them @xmath44 in grey ( green ) within 0.4 time units , for the attractors of @xmath108 ( top ) and of @xmath113 ( bottom ) .
the vertical scales are the same as in fig .
[ fig : a3s4intime ] .
@xmath85 and @xmath76 are plotted as @xmath140 .
thus the @xmath85- and @xmath76-zeros appear as logarithmic singularities.,scaledwidth=80.0% ] the global structure of the @xmath2 parameter space , as exhibited in fig .
[ fig : rsigma2000 ] , is obviously related to the existence and stability of such orbits , with winding numbers @xmath141 increasing as @xmath0 and @xmath1 ( and thereby @xmath101 ) grow .
the observations to be explained in the subsequent theoretical section are the following : 1 . repetition of bands : along the line @xmath142 with a fixed ratio @xmath143 , there exists a period @xmath144 in @xmath0 ( and consequently @xmath145 in @xmath1 ) such that if an attractor of type a@xmath100 ( or s@xmath100 ) exists for values @xmath146 , then a corresponding attractor of type a@xmath147 ( or s@xmath147 ) exists for @xmath148 .
@xmath143-dependence of @xmath0-periods @xmath144 : from the shape of the bands it may be inferred that @xmath149 has a minimum near @xmath150 .
this means , the periodic repetition of the bands is fastest ( in @xmath0 ) along the line @xmath151 .
size scaling of dominant orbits with @xmath0 : the size of these orbits , i.e. , the radius @xmath152 of their @xmath75-projections measured in units of @xmath0 , is independent of @xmath0 : @xmath153 for fixed @xmath143 .
thus far we have not addressed the question of how @xmath152 depends on @xmath143 , so let us see next what theory has to tell ; the result is presented in fig .
[ fig : rhomegavsm ] where @xmath154 is plotted as a function of @xmath143 .
of course , the above statements only hold for large @xmath0 and @xmath1 , in practice for @xmath155 and @xmath1 in the bands above @xmath156 .
we begin with a discussion of appropriate scalings of the entire set of lorenz equations ( [ lorenz ] ) when both parameters @xmath0 and @xmath1 become large .
their asymptotic behavior with @xmath157 depends on how @xmath1 is assumed to change simultaneously .
robbins in @xcite and also sparrow @xcite considered the case @xmath158 . here
we study the behavior of the lorenz equations with prandtl numbers increasing as @xmath159 , with some fixed exponent @xmath160 to guarantee increasing @xmath1 , but @xmath161 for @xmath1 to stay below the upper stability border @xmath162 .
if r and @xmath1 become very large , the motion in @xmath85 and @xmath76 is expected to become very fast because @xmath163 in the equations of motion becomes large .
thus we introduce a rescaled , stretched time @xmath164 by defining @xmath165 with a scaling factor @xmath166 , @xmath167 .
let us now also rescale the variables @xmath85 and @xmath76 which measure , respectively , the amplitudes of the role velocity and temperature . as for the rescaling of
@xmath44 a glance on the second lorenz equation shows that @xmath76 is driven by @xmath168 . this fact is reflected in the amplitudes @xmath44 of the numerical trajectories , or in the behavior of the fixed points p@xmath59 , both as functions of @xmath0 .
thus we define [ rescaling ] x = , y = , z - r = . the lorenz equations then read , with @xmath169 etc . , [ lorenzrescaled ] =-+ , =-- , =-b ( + ) + .
the freedom of choosing scaling parameters may now be used to normalize to unity the coefficients of the driving term in the first equation , and of the two coupling terms @xmath170 and @xmath171 , which leads to [ scalerelations ] = , = , = .
these equations allow us to express @xmath172 , @xmath173 , and @xmath174 in terms of @xmath1 and @xmath175 .
we find first @xmath176 , thus @xmath177 , then @xmath178 , and finally @xmath179 .
the @xmath180 signs may be interpreted as corresponding to the symmetry @xmath45 which is clearly preserved under the rescaling .
we can therefore restrict ourselves to the plus sign and have [ factors ] = ^-1r^e , = = ^-2 ^ -1r^2e - s .
the asymptotic behavior will depend on how the exponent @xmath181 compares to the other exponent @xmath182 .
there appear to be three distinct possibilities .
one is that @xmath173 and @xmath174 grow faster than @xmath172 , i.e. , @xmath183 or @xmath184 .
the second is that all three variables @xmath185 scale with the same factor , @xmath186 or @xmath187 .
the third possibility @xmath188 may be excluded with the argument that then the relaxation term @xmath189 in the @xmath190-equation would diverge .
a representative case of the first possibility is @xmath191 and @xmath192 , meaning @xmath193 and @xmath194 .
this was the choice of robbins @xcite and sparrow @xcite .
all dissipative terms vanish in this limit , and the equations reduce to [ sparrow ]
= , = - , = , (= , = 1/ ) .
this system has the two integrals @xmath195 and @xmath196 ; it is equivalent to the equations of a pendulum with length @xmath197 and energy @xmath198 .
the values of the integrals @xmath198 and @xmath197 have to be determined from the @xmath175-correction to ( [ sparrow ] ) , i.e. , by considering their slow dynamics . in @xcite it
was shown how this can be done in terms of elliptic integrals . in that sense , this case has been solved .
let us therefore concentrate on the second possibility @xmath187 .
then @xmath199 is independent of @xmath0 , and we can take @xmath200 .
all scaling factors can now be expressed in terms of the prandtl number @xmath1 .
the term @xmath201 in the second lorenz equation may be neglected as vanishingly small in comparison to with @xmath202 .
hence the rescaled eqs .
( [ lorenzrescaled ] ) become [ lorenzrescaled2 ] =-+ , =- ,
= - - , . but what about the two last terms in the @xmath203-equation ? their order of magnitude is @xmath204 and @xmath205 , respectively . again
, there are three possibilities : + case a. with the choice @xmath206 , or @xmath207 , the second term will tend to a constant and dominate the decreasing first term .
we can write @xmath208 with some constant @xmath209 .
the set of equations becomes [ case1/2 ] =-+ , =- , =-w^-2 .
when @xmath210 varies between @xmath211 and @xmath27 , the same equations apply because the last term in the @xmath203-equation of ( [ lorenzrescaled2 ] ) dominates the second .
note that no comparison can be made between these two terms and @xmath212 .
+ case b. when @xmath213 , the two last terms in the @xmath203-equation are of the same order of magnitude @xmath214 .
we write @xmath215 and obtain the set of equations [ case1 ] =-+ , =- , =-(+ ) .
case c. when @xmath216 , @xmath1 grows faster than @xmath0 , so we leave the range where @xmath217 .
formally it is found that both terms besides @xmath212 in the @xmath203-equation may be neglected , but in contrast to eqs .
( [ sparrow ] ) there remains a dissipative term in the @xmath190-equation .
the constant of motion @xmath218 constrains the solution to a cylinder . on this cylinder
the equation reduces to the damped pendulum , and therefore almost all initial conditions decay towards the stable equilibrium on the respective cylinder .
we thus have to discuss the asymptotic regimes a and b , and do this separately . to our knowledge , the two sets of eqs .
( [ case1/2 ] ) and ( [ case1 ] ) have not been analyzed so far . as we shall see , their solutions lead to an understanding of the phase space structure displayed in fig .
[ fig : rsigma2000 ] and of the numerical findings in section [ sec : attractors ] .
we start with the simpler case of the asymptotic lorenz equations , the set of the rescaled eqs .
( [ case1/2 ] ) . of all the complex dynamics that it contains , we only want to describe periodic orbits of the simplest kind . from our numerical studies
we anticipate that these orbits have two clearly distinct phases .
there is slow motion down along the @xmath44-axis , and a fast return to the original large value of @xmath44 via a large loop away from the @xmath44-axis . the fast motion is readily described when we neglect the constant term in the @xmath203-equation of ( [ case1/2 ] ) .
then , combining the equations for @xmath219 and @xmath203 , we obtain @xmath220 which after integration gives the circles [ eqn : etazetaloop ] ^2 + ^2 = ^2 with as yet undetermined radius @xmath221 .
a convenient parametrization of these circles is found with a rescaling of time , defining @xmath210 as an arc length via @xmath222 .
the equations for @xmath223 and @xmath224 are thereby linearized , with solutions [ eqn : etazetaofs ] ( s ) = s , ( s ) = - s .
this explains the observation that @xmath203 oscillates between two equal values of different sign , namely @xmath225 .
the equation for @xmath190 , [ eqn : loopxi ] = -(- ) = - ( 1- ) does not appear to be analytically solvable , but as long as @xmath190 is not too small , it describes a relaxation of @xmath226 towards @xmath223 .
hence , with a delay in @xmath164 of the order of @xmath27 , @xmath190 follows @xmath219 on its excursion to values of the order of @xmath221 .
consider now the slow motion .
it is given by the exact dynamics on the stable manifold @xmath227 of the fixed point @xmath228 .
the solution with initial point @xmath229 is @xmath230 and reaches the lower value @xmath231 at time @xmath232 .
linearizing the set of equations ( [ case1/2 ] ) around this motion , we obtain
= - + , = -(-
the @xmath164-dependent eigenvalues are @xmath233 with discriminant @xmath234 .
three @xmath164-regimes may be identified .
assuming @xmath235 , the discriminant is initially negative ; the motion is oscillatory and decaying according to @xmath236 .
this regime ends at the time @xmath237 defined by @xmath238 , or @xmath239 . in the second regime , the motion continues to be decaying but is no longer oscillatory ; this is the case between @xmath237 and @xmath240 where @xmath241 .
the third regime , between @xmath242 and @xmath243 , sees @xmath190 and @xmath219 increasing at a growing rate . to proceed with the @xmath244-equations
, we introduce a new variable @xmath245 by separating the exponentially decaying part from @xmath190 . with @xmath246 ,
the two linear equations of first order are transformed into [ eqn : xai ] x - ( ) x = 0 , x - x = 0 where @xmath247 , and @xmath248 lives on the interval @xmath249 .
this equation is the well known airy equation .
its general solution is a superposition of the two airy functions @xmath250 and @xmath251 where @xmath252 .
both solutions oscillate for @xmath253 . for @xmath254
the solution ai decays whereas bi grows exponentially .
we are interested in the solution @xmath255 , or [ eqn : xiaisol ] ( ) = ^-_1/2 ^ -/2(w^-2/3 ) . with @xmath256 .
it may be checked that @xmath257 is governed by the same airy equation and thus , up to a phase shift , has the same solution .
an example of the function bi is shown in the left part of fig .
[ fig : airyplots ] , where the parameters are those of s@xmath98 ( the plots are made on the basis of eqs .
( [ eqn : xisol])-([eq : rhovsm ] ) ) .
the corresponding @xmath258 is shown in the right part of the picture , in logarithmic representation .
from here it is straightforward to determine the value of @xmath221 , and the periodicity of the pattern of bands . we know from the analysis of the outer part of the attractor , see ( [ eqn : etazetaofs ] ) , that the solution starts , for @xmath259 or @xmath260 , at about the same values @xmath190 and @xmath219 at which it ends for @xmath261 : [ rhoequation ] ( 0 ) = ( t )
. using the asymptotic behavior of bi for large negative and positive arguments , [ airyasymp ] ( -|z| ) & ( |z|^3/2 + ) , + ( |z| ) & \ { |z|^3/2 } , and the fact that the exponential terms dominate the algebraic factors by orders of magnitude , we simply equate the exponents on the two sides of ( [ rhoequation ] ) . for @xmath260
( [ eqn : xiaisol ] ) gives @xmath262 ( zero exponent ) ; for @xmath263 , we add up the exponents in eqs .
( [ eqn : xiaisol ] ) and ( [ airyasymp ] ) , with @xmath264 , and obtain @xmath265 inserting @xmath232 and @xmath266 we obtain [ rhovalue ] 144 ^2 = ( 1 + 4)^3 . the relevant solution is @xmath267 . to determine the values @xmath268 where the orbits have a certain number @xmath125 of @xmath190- and @xmath219-zeros at each close encounter with the @xmath44-axis , we use the asymptotic behavior of the airy function @xmath269 as given in ( [ airyasymp ] ) .
consider first the symmetric orbits s@xmath100 as they develop from time @xmath270 to @xmath271 .
the coordinates @xmath190 and @xmath219 come in from large values ( positive or negative ) and go out to large values of the opposite sign .
this means the phase shift between @xmath272 and @xmath273 is @xmath274 ; the cosine performs an odd number of half oscillations . for the asymmetric orbits a@xmath100 ,
on the other hand , the phase shift is a multiple of @xmath275 .
taken together , this leads to the `` quantum condition '' [ wn1 ] 23_1 ^ 3/2 + = n , where @xmath123 corresponds to the first symmetric and @xmath276 to the first asymmetric band . with @xmath277
we get [ wn ] w_n = 1.02 , n = 1,2 , ... . to wrap it up
, the results for this regime of large @xmath0 are the following .
the @xmath125-th band is centered around the line @xmath278 ; the size of the corresponding attracting periodic orbits in the original variables @xmath279 is @xmath280 . but note this is only asymptotically true and can hardly be seen in the diagram of fig .
[ fig : rsigma2000 ] except in the lower right corner where the s@xmath71-band is just bending over to follow the @xmath281 behavior .
@xmath282 gives the prediction @xmath283 which lies at the lower border of the s@xmath71-band .
the bands corresponding to higher values of @xmath125 have not yet entered this asymptotic regime .
let us now study the set of equations ( [ case1 ] ) @xmath284 where @xmath285 is the constant parameter . much of the analysis is similar as in the previous subsection , but the explicit occurrence of @xmath0 in the equations adds complications .
the general strategy is the same . for the outer , large @xmath190 and @xmath219 parts of the solution we neglect the term @xmath286 in the @xmath203-equation and get identical results as in ( [ eqn : etazetaloop])-([eqn : loopxi ] ) .
the slow motion along the @xmath44-axis is more complicated even though the @xmath203-equation , neglecting the @xmath287-term , decouples again from the rest : its solution is no longer linear in @xmath164 . with the initial condition @xmath229 the approximate @xmath288-equation
is solved by [ zetaoftau ] ( ) = ( + 1m)^-(b / mr ) - 1 m , and the time @xmath243 needed to evolve from @xmath221 to @xmath231 is [ eqn : t ] t = .
using this slow exponential decrease of @xmath203 ( with the rate @xmath286 ) in the equation @xmath289 , we have again a time dependent linear system of equations for @xmath190 and @xmath219 .
its eigenvalues at fixed time are [ xietaeigenvalue ] = - = - ( ) = + 1 m - ( + 1 m ) ^-(b / mr ) . the discriminant @xmath290 is negative in the @xmath164-range @xmath291 where @xmath237 is defined by the condition @xmath292 : [ eqn : tau1 ] _ 1 = . in this @xmath164-range
the amplitudes @xmath293 oscillate and decay with damping rate @xmath294 .
the second regime ( further exponential decrease of @xmath190 and @xmath219 , but now without oscillations ) ends at @xmath242 , defined by the condition that one of the eigenvalues @xmath295 be zero , i.e. , @xmath241 : [ eqn : tau2 ] _ 2 = ( 1 + m ) . from @xmath242
to @xmath243 one eigenvalue is positive and the variables @xmath190 and @xmath219 increase at a growing rate .
this behavior is in complete qualitative agreement with fig .
[ fig : a3s4logintime ] . with the ansatz @xmath246 , the two linear equations of first order are transformed into @xmath296 which is , in contrast to the previous case a , not the airy equation because it is not linear in @xmath164 .
however , with the substitution @xmath297 via @xmath298 it may be reduced to the standard form of the bessel equation .
the @xmath293 -solution is then a linear combination of the two bessel functions of first and second kind of order @xmath299 and argument @xmath300 . for fixed ,
finite @xmath164 but large @xmath0 we have @xmath301 , linear in @xmath164 , but with a large additive constant .
thus in the limit of large @xmath0 the asymptotics for bessel functions with large negative order and large arguments is needed .
the simplest way to obtain such asymptotics is to use an approximation by airy functions from the very beginning when solving the @xmath245-equation .
this is done by linearizing the discriminant @xmath302 in the original equation at an appropriate time @xmath303 , [ deltalin ] ( ) = ( ^ ) + ( ( ^ ) + ) ( - ^ ) + , upon which we obtain the airy equation for @xmath304 , [ eqn : x ] - x = 0 , = ( ( ^)+ ) , where [ eq : tauast ] = - _ 1^ _ 1^= ^- ( ^)/ . but what should we take for @xmath303 ?
a natural choice seems to be @xmath305 where @xmath306 and @xmath307 , whence [ alpha1 ] = _ 1 = ( 1 + ) _ 1^= _ 1 .
however , if we care to treat the decaying and the exploding part of the @xmath293 -dynamics with equal weight , then it appears natural to take the minimum of @xmath246 as the reference point , i.e. , @xmath308 . here
@xmath241 and @xmath309 , so that from and we obtain [ alpha2 ] = _ 2 = _ 1^= _ 2- = _ 1 + o((m/4)^2 ) .
the last step involves @xmath310 and an expansion of @xmath311 to second order in the small quantity @xmath312 .
the difference @xmath313 is itself of second order .
finally , when we are mainly interested in the initial phase of the decay , @xmath314 ought to be a reasonable choice . then @xmath315 and @xmath316 which leads to [ alpha0 ] = _ 0 = ( 1 + m ) _ 1^ = _ 1 + o((m/4)^2 ) .
again @xmath313 is of second order in small quantities . assuming that the @xmath143-corrections are not too large , we take @xmath317 in all three cases and find the solution [ eqn : xisol ] ( ) = ^-_1/2 ^ -/2(^1/3 ) , with @xmath256 .
an example of this time development is given in fig .
[ fig : airyplots ] where the parameters corresponding to s@xmath98 are taken .
the left picture shows bi(@xmath318 ) in the time range @xmath319 , the right picture is a logarithmic plot of @xmath258 for an entire period . the qualitative agreement with the numerical integration of the lorenz equations ,
see fig .
[ fig : a3s4logintime ] , is quite impressive .
plots of @xmath320 ( left ) and @xmath321 ( right ) , where @xmath322 varies on the ( @xmath0-independent ) interval @xmath323 $ ] .
the parameters are @xmath324 , corresponding to s@xmath98.,scaledwidth=80.0% ] as for the value of @xmath172 , we take @xmath325 to determine @xmath221 ( a glance at fig .
[ fig : a3s4logintime ] shows that the gross feature of the time development is nearly symmetric with respect to @xmath133 ) , and @xmath326 or @xmath327 for the periodicity of the oscillations in the initial phase of the decay ( which takes place between @xmath259 and @xmath328 ) .
this is why the @xmath172 s are different in the left and right part of fig .
[ fig : airyplots ] .
the right part shows that the periodicity condition @xmath329 , see ( [ rhoequation ] ) , is approximately fulfilled for @xmath330 if we take @xmath331 . the number of zeros , however , is not correct ( 6 instead of 7 ) ; had we taken @xmath332 , the values of @xmath333 would have been different between beginning and end . on the other hand ,
the left part shows that with @xmath332 we get the correct number of zeros ( the same is true for @xmath334 ) .
this reflects the fact that @xmath302 is not a linear function of @xmath164 . using the asymptotics of bi
, the periodicity condition @xmath329 implies @xmath335 this leads to the following equation for the dependence of @xmath221 on @xmath143 , [ eq : rhovsm ] = ( ) ^3/2 . the left part of fig .
[ fig : rhomegavsm ] is a plot of the corresponding function @xmath336 ; the dots are derived from numerical test values of @xmath203 that were generated by integrating the full lorenz equations .
the agreement is satisfactory . in the limit of small @xmath143 , obtained by expanding to lowest order in @xmath337 and @xmath338
, we recover the result ( [ rhovalue ] ) . with increasing @xmath143 , the amplitude @xmath221 of the @xmath203-motion decreases from the value @xmath339 to 0.493 at @xmath340 . for @xmath143 of the order of 1
the analysis becomes less and less reliable as the linearization of @xmath302 , and hence the use of airy functions , ceases to be a good approximation .
when the value @xmath341 is compared to the numerically determined value of @xmath152 , see e.g.figs .
[ fig : attractora3 ] and [ fig : attractors4 ] , we must remember that @xmath152 was scaled with @xmath0 whereas @xmath221 is scaled with @xmath215 ; hence @xmath342 , and @xmath343 which agrees well with the numerical observations .
left : the amplitude @xmath344 of the @xmath203-motion as given by .
the points are obtained by numerical integration of the full lorenz equations .
right : the frequency @xmath345 of the repetitive pattern in the parameter @xmath0 , as a function of @xmath143 . from top to bottom , the curves are calculated with @xmath346 , @xmath347 , and @xmath348 as defined in the text and inserted into ( [ eq : rnfull ] , [ eq : pm ] ) , with @xmath336 taken from in all three cases .
, scaledwidth=80.0% ] to determine the periodicity of the @xmath2-phase space pattern as we go along a line @xmath142 with fixed @xmath143 in fig .
[ fig : rsigma2000 ] , the same arguments as in the derivation of lead to 23_1 ^ 3/2 + = n . but
again the question arises as to which @xmath172 should be taken .
since @xmath221 was calculated with @xmath325 it seems consistent to do the same here ; the result is the lower of the three curves in the right part of fig .
[ fig : rhomegavsm ] .
it might be argued , however , that the behavior of the solution between @xmath259 and @xmath237 is rather determined by the slope of @xmath302 in that range , so we took @xmath326 and @xmath327 for comparison .
the three results for the @xmath0-values of the characteristic points a@xmath100 and s@xmath100 cf .
( [ windingnumbers ] ) for the relation between @xmath101 and the winding number @xmath125 may be summarized as follows : [ eq : rnfull ] r_n(m ) = ( n - 14)p(m ) with half period ( in @xmath0 ) [ eq : pm ] p(m ) = ( ) ^-3/2 , where @xmath349 corresponds to @xmath350 , namely @xmath351 , @xmath352 , and @xmath353 .
in the regime of low @xmath143 , all three choices of @xmath172 give the same frequency @xmath354 . with @xmath277 and the expansion of the @xmath355 term we recover the result from the previous subsection : @xmath356 with @xmath357 given by eq .
( [ wn ] ) . for
larger @xmath143 the best agreement with the numerical integration of the original lorenz equations is obtained from @xmath351 : with the choice @xmath150 the prediction for the period of the pattern is @xmath358 .
this is what was used to define the points s@xmath100 and a@xmath100 in fig .
[ fig : rsigma2000 ] , namely , @xmath359 .
if instead of @xmath360 we take @xmath361 in ( [ eq : rnfull ] ) , we obtain values @xmath362 corresponding to the lower border of the @xmath0-bands .
these are plotted in fig .
[ fig : patterntest ] , for @xmath125 from 1 to 9 and parameters @xmath143 ranging from 0 to approximately 1.25 .
the agreement with fig .
[ fig : rsigma2000 ] is qualitatively convincing and quantitatively satisfactory .
the curves @xmath363 for @xmath364 and @xmath365 .
the dashed lines are @xmath366 and @xmath367 .
, scaledwidth=60.0% ]
while the periodic orbits discussed in the previous two sections form the backbone of the structure in the @xmath2-phase diagram on large scales of @xmath0 and @xmath1 , they do not represent the whole picture . in this section
we report on the scheme of bifurcations that connects the different bands , and we also give a short discussion of the coexistence of attractors .
in addition to the periodic attractors of periods 1 and 2 , there are small bands of stability for periodic orbits of all periods , and of chaotic motion .
this is summarized , using a color code , in figs .
[ fig : r409sigma60 ] and [ fig : rsigma2000 ] . within a given band ,
the attractors are the same type , i.e. , they share the same period and symmetry property with respect to the reflection @xmath45 , see ( [ symmetry1 ] ) .
we use the notation `` ps '' for p - periodic symmetric attractors , and `` pa '' for p - periodic asymmetric attractors ( whose mirror images under @xmath45 are also attractors ) .
@xmath368-bifurcation diagram at fixed @xmath369 , @xmath370 for @xmath371 .
the initial conditions are @xmath372 ( red ) and @xmath373 ( green ) .
attractors are asymmetric when they carry only one of the colors red or green .
there are also symmetric attractors , given in black ( superposition of red and green , as they are reached from both initial conditions ) .
a few periodic orbits are shown in @xmath75-projection , of periods 2 in the upper right part , and periods 4 in the lower right ; arrows indicate the corresponding branches of the bifurcation diagram .
the conspicuous merging of the two major chaotic bands , as @xmath1 is lowered , seems to occur at or very close to the value @xmath374 . ]
as the structural features of the @xmath2-phase diagram are different in the regions of low and high @xmath1 ( separated roughly by @xmath375 ) , we discuss them in turn . to begin with , we present in fig .
[ sigmaxbifdia ] a survey on the bifurcation behavior of attractors along the line @xmath369 for @xmath371 .
this @xmath1-range bridges the two regions of small and large prandtl numbers , respectively .
it starts well below @xmath376 ( @xmath377 for this @xmath0 ) and ends far above .
the bifurcation scheme along this line is shown in the left part of fig .
[ sigmaxbifdia ] .
it records the @xmath85-values of the attractors in the poincar surface of section @xmath87 .
red points were computed with initial conditions @xmath372 , green points with @xmath378 .
when the colors do not mix , they represent asymmetric attractors ; symmetric attractors are reached from both initial conditions and appear in color black . starting with the highest @xmath1 , we observe a small black segment of symmetric orbits with period 2 , i.e. , the attractors are type 2s .
the @xmath75-projection for @xmath379 is shown at the right side .
lowering @xmath1 to approximately 62 , the 2s - orbit splits into two 2a - orbits , one red , the other green , mirror images of each other under @xmath45 .
this is illustrated for @xmath380 .
each of these two attractors undergoes a period doubling scenario , 2a @xmath381 4a @xmath381 8a @xmath381 , corresponding to red , green , blue bands in fig .
[ fig : r409sigma60 ] , until chaos appears at about @xmath382 .
when @xmath1 is further decreased , the inverse cascade of band merging as described in @xcite is observed .
it ends at @xmath383 , where the two bands of asymmetric chaos merge into one band of symmetric chaos , thereby restoring the symmetry under @xmath45 . as @xmath1 falls below approximately 41.2 , a period 4 window emerges by an intermittency scenario . at first
the attractor is symmetric , of type 4s , as shown for @xmath384 .
again , this orbit undergoes a pitchfork bifurcation into two asymmetric orbits of period 4 , shown in red and green for @xmath385 , and then the period doubling sequence takes over for both of them . in fig .
[ fig : r409sigma60 ] , this window appears as the dominant green band in the white chaotic region .
blowups of two pieces of the upper part of fig .
[ sigmaxbifdia ] . left : @xmath386 , upper left branch ; right : @xmath387 , upper right branch .
the two pictures are like mirror images of each other .
again , red and green indicate the asymmetric attractors , approached from the two @xmath45-images of the initial values .
, scaledwidth=90.0% ] fig .
[ fig : blowups ] shows the upper part of the bifurcation diagram in fig .
[ sigmaxbifdia ] in greater detail , for the intervals @xmath387 and @xmath388 . except for their different scales
the pictures are much like mirror images of each other . in both cases
the scenarios coincide in many aspects ( though not completely ) with the bifurcation scheme of the antisymmetric cubic map x_n+1 = f(x_n ) = ( 1-c)x_n + cx_n^3 , -1 x 1 , [ eq : cubic ] shown in fig .
[ fig : cubic ] in the ranges @xmath389 and @xmath390 .
this similarity was pointed out and explored using symbolic dynamics in @xcite and @xcite .
the reason for the good correspondence seems to be that is the simplest polynomial 1-dim map that shares with ( [ lorenz ] ) a reflection symmetry .
the invariance with respect to @xmath45 is here represented by the symmetry under @xmath391 , namely @xmath392 .
this modifies the standard period doubling scenario of the quadratic map @xmath47 in such a way that the first bifurcation of a symmetric orbit of period p is not a period doubling ps @xmath381 ( 2p)s but rather a pitchfork bifurcation where ps breaks up into pa and its mirror image @xmath393 . only then does period doubling start : pa @xmath381 ( 2p)a @xmath381 ( 4p)a @xmath381 .
let us now consider what kind of regularity we can detect in the region of large @xmath0 and @xmath1 , in the onion like structure of fig . [ fig : rsigma2000 ] .
a representative picture of the bifurcation behavior is obtained by considering any parameter interval that connects two consecutive bands of equal symmetry .
we take the line from a@xmath71 to a@xmath19 and show the corresponding @xmath396-bifurcation diagram in the upper left part of fig .
[ fig : rxbifdia ] , with the two initial conditions @xmath397 ( red ) and @xmath398 ( green ) ; @xmath399 in both cases .
( this large value of @xmath400 is to a large extent arbitrary ; however , it was observed that chaos coexists with periodic orbits , and that initial conditions with @xmath400 of order @xmath0 or lower tend to be captured by the latter . )
the poincar surface of section is @xmath401 as usual .
only the branch with @xmath402 is shown : the complete picture contains the mirror image @xmath403 , with colors red and blue interchanged .
the small windows in fig . [ fig : rxbifdia ] present a number of characteristic attractors in @xmath75-projection , arrows connecting to the respective points on the bifurcation diagram .
on the upper branch starting from a@xmath19 , we have asymmetric periodic attractors of type 1a .
these undergo a period doubling scenario ( orbit 2a , etc . ) , including the inverse cascade ( with asymmetric chaotic orbits ) , before they merge with their image under @xmath45 and yield two symmetric chaotic orbits .
this scenario is illustrated with the orbits in the upper right part of the picture . upon lowering @xmath0 ,
the chaotic attractor suddenly disappears , and an orbit of type 2s emerges , characteristic of the red band with center s@xmath19 .
this symmetric orbit splits into two asymmetric orbits of type 2a , in a pitchfork bifurcation like in fig .
[ sigmaxbifdia ] ( whose rightmost part belongs to the band s@xmath71 rather than s@xmath19 ) .
next , period doubling sets in again .
the chaos in the range of inverse cascading is at first asymmetric ( though different from the one in the yellow band ) , but turns into symmetric chaos at the end point of the band merging .
finally , an attractor of type 1a emerges from this chaos and gives rise to a picture around a@xmath71 similar to the one with which we started at a@xmath19 .
we have not investigated the mechanism by which the `` new '' stable periodic orbits are generated when the chaotic motion becomes unstable upon lowering @xmath0 .
in fact , they coexist with the chaotic attractors before they are seen in fig .
[ fig : rsigma2000 ]
. we can not tell whether they emerge in an intermittency scenario .
it may be that their range of existence , ignoring the question of their stability , extends all the way up to infinity .
upper left : @xmath396-bifurcation diagram with @xmath404 and @xmath405 , @xmath370 .
the initial conditions are @xmath406 and @xmath93 ( red ) or @xmath407 ( green ) .
the 8 small windows show attractors corresponding to the parameter values identified by the arrows , in @xmath75-projection . following the sequence in clockwise direction , starting at the top ,
the bifurcation scheme between a@xmath19 ( @xmath408 ) and a@xmath71 ( @xmath409 ) is illustrated : 1a - orbits ( yellow band with center a@xmath19 , see fig .
[ fig : rsigma2000 ] ) undergo period doubling , then become chaotic . out of this chaos ,
2s - orbits emerge ( red band with center s@xmath19 ) and undergo a pitchfork bifurcation into 2a and its @xmath45-image ; these bifurcate according to the period doubling scenario before chaos sets in , first asymmetric , then symmetric .
finally , 1a - orbits ( of the a@xmath71-band ) emerge and the scenario starts anew . ] fig .
[ fig : rhocheck ] is an extended version of the bifurcation diagram in fig .
[ fig : rxbifdia ] .
the range of @xmath0 is from 50 to 2050 , and instead of the ordinate @xmath85 we plot @xmath410 , again for @xmath87 .
remember that for a given periodic orbit , @xmath76 reaches its maximum value when @xmath44 is near @xmath0 .
the picture contains two messages : ( i ) the repetitive nature of the pattern in fig .
[ fig : rsigma2000 ] extends to the bifurcation scenarios , and ( ii ) the value of @xmath410 at its maximum , for periodic attractors , seems to tend to a constant @xmath411 as @xmath412 and @xmath413 .
this was an essential assumption in the scaling scheme that we used in the analysis of section [ ssecm ] .
the value @xmath414 that our theoretical investigation gave for @xmath150 is close to the observed limit .
it has already been mentioned several times that different attractors may coexist .
for example , whenever an attractor a is not in itself symmetric under @xmath45 , it coexists with its partner @xmath45a .
this applies in particular to the fixed point attractors p@xmath59 in the black region , or to the 1a - attractors in the yellow bands .
consequently , there must be at least four coexisting attractors where the yellow bands extend into the black region , as can be seen in fig .
[ fig : rsigma2000 ] .
a less trivial case is analyzed in fig .
[ fig:4basins ] .
the parameters @xmath417 are located in the lower part of fig .
[ fig : r409sigma60 ] where the green asymmetric period 4 band overlaps with the black region .
the left part of fig .
[ fig:4basins ] shows the poincar surface of section @xmath87 , with initial conditions @xmath418 colored pixel by pixel according to the final state to which they are attracted : white or blue if the final state is the fixed point p@xmath419 , respectively , red or gray if it is one of the two 4a - orbits .
no other attractor besides these four was found .
the blue and white basins of the point attractors p@xmath59 appear to have smooth boundaries whereas the boundary between the basins of attraction of the two 4a - orbits is highly entangled .
left : basins of attraction of 4 coexisting attractors , for @xmath417 .
initial conditions in the @xmath418-plane , @xmath420 , with @xmath80 , are scanned pixel by pixel and given a color depending on the attractor to which they are asymptotic : white for p@xmath421 , blue for p@xmath422 , red or gray for the asymmetric period 4 orbits shown on the right ( abscissa @xmath76 and ordinate @xmath44 ) .
the fixed point attractors p@xmath59 are marked as black dots .
all intersection points ( 8 each ) of the two period 4 attractors with the poincar surface @xmath84 are emphasized . ]
another case of coexisting attractors is documented in fig .
[ fig:3basins ] where the parameters @xmath423 are from the chaotic band at the upper rim of the lowest yellow band in fig .
[ fig : rsigma2000 ] . fig . [ fig:3basins ] demonstrates how a symmetric chaotic attractor coexists with two attractive asymmetric period 1 orbits .
no other attractors have been detected .
the basin boundaries seem to be complicated but not fractal ; they are spiralling out from the unstable fixed points p@xmath59 .
each point is colored according to which attractor absorbs it .
each basin trivially contains its respective attractor .
the location of their intersections with the plane @xmath87 is marked by black dots .
left : basins of attraction of three coexisting attractors for @xmath423 .
initial conditions with @xmath424 , @xmath80 are scanned and colored according to their final fate .
the three attractors are shown on the right , in @xmath75-projection : two asymmetric orbits of period 1 ( with the blue and green basins of attraction ) , and one symmetric chaotic attractor ( in black , with the white basin of attraction ) .
the intersections of the attractors with the poincar surface are emphasized by arrows . ]
inspection of figs .
[ fig:4basins ] and [ fig:3basins ] suggests that a scan of initial conditions along the line @xmath425 , @xmath426 , might catch all ( co-)existing attractors . thus instead of scanning a two - dimensional set of initial conditions , we scan along this diagonal line only and use the second dimension to superimpose these scans for a range of parameters @xmath1 .
such a `` @xmath1-stack '' is shown in fig .
[ fig : attractorcoexistence ] for @xmath427 .
the abscissa is @xmath425 in the range from @xmath428 to @xmath429 , and the ordinate is @xmath1 from @xmath428 to @xmath430 .
comparison with fig .
[ fig : rsigma2000 ] shows that this @xmath1-range covers the bands containing the points s@xmath71 , a@xmath71 , and s@xmath19 , in addition to the low-@xmath1 structure with regular behavior at the very bottom and chaos in a range around @xmath431 .
the coloring scheme is somewhat arbitrary and was designed so that the coloring of different attractors could be done automatically . for a fixed value of @xmath1
different color designates different attractors , i.e. , the number of different colors on a scan - line is the number of coexisting attractors .
same color for adjacent points with different @xmath1 implies that the corresponding attractors are deformations of each other .
color white is reserved for chaotic motion .
let us concentrate on the @xmath1-ranges which are marked as s@xmath432 , s@xmath433 , c@xmath71 , etc .
s@xmath433 is the @xmath1-interval @xmath434 .
the attractors are symmetric orbits of period 2 and type s@xmath71 , see fig .
[ fig : winding ] , and there are no other attractors ; hence there is only one color . for smaller @xmath1 in the range
s@xmath432 , @xmath435 , the attractors have undergone a pitchfork bifurcation and exist in pairs of asymmetric orbits of period 2 .
thus we see two colors , corresponding to the two basins of attraction . at larger
@xmath1 , on the other hand , in the range @xmath436 , the attractors of type 2s coexist with chaotic attractors ; again we observe two colors . the band with orbits of type a@xmath71 exists in the range @xmath437 .
as the attractors are asymmetric orbits , they occur in pairs , and two colors appear ( black and magenta ) . at lower @xmath1 , we have a range @xmath438 where period doubling has increased the period but not the number of attractors ; thus there are again two colors . at larger
@xmath1 , in the range @xmath439 , the two asymmetric periodic attractors coexist with a chaotic attractor , and three colors are needed .
[ fig:3basins ] is an example from this range .
if @xmath1 is further increased , we enter the regime which is discussed in fig .
[ fig : rxbifdia ] .
the bifurcation tree of the s@xmath19-band leaves a complicated trace in fig .
[ fig : attractorcoexistence ] , but then there comes a range where a 2s - orbit is the system s only attractor . from thereon , the pattern seems to repeat itself , but the details on the @xmath440 axis become more and more intricate .
@xmath1-stack of domains of attraction , for @xmath427 . for @xmath1 in the range between @xmath428 and @xmath430 ( ordinate ) initial points along the diagonal @xmath441 ( and in the poincar plane @xmath87 ) are given a color according to which orbit attracts them .
white , in particular , means there are chaotic orbits .
the values @xmath441 are taken from the range @xmath442 .
the features of the @xmath1-ranges s@xmath432 , s@xmath433 , c@xmath71 , etc . are explained in the text . ]
ding and b .-
systematics of the periodic windows in the lorenz model and its relation with the antisymmetric cubic map . , 9:375 , 1988 .
reprinted in _ chaos ii _
hao , world scientific , singapore ( 1990 ) . | the parameter dependence of the various attractive solutions of the three variable nonlinear lorenz model equations for thermal convection in rayleigh - bnard flow is studied .
its bifurcation structure has commonly been investigated as a function of @xmath0 , the normalized rayleigh number , at fixed prandtl number @xmath1 .
the present work extends the analysis to the entire @xmath2 parameter plane .
an onion like periodic pattern is found which is due to the alternating stability of symmetric and non - symmetric periodic orbits .
this periodic pattern is explained by considering non - trivial limits of large @xmath0 and @xmath1 .
in addition to the limit which was previously analyzed by sparrow , we identify two more distinct asymptotic regimes in which either @xmath3 or @xmath4 is constant . in both limits the dynamics is approximately described by airy functions whence the periodicity in parameter space can be calculated analytically .
furthermore , some observations about sequences of bifurcations and coexistence of attractors , periodic as well as chaotic , are reported . |
decay of a heavy particle with a photon in the final state can often lead to experimental signals for physics beyond the standard model .
recent literature contains many studies , both experimental and theoretical , devoted to signals of the type @xmath2 + @xmath3 , @xmath4 + @xmath3 etc .
@xcite in the context of various new physics scenarios . a rather large fraction of such studies
are concerned with supersymmetric ( susy ) @xcite theories .
for example , in models with gauge mediated supersymmetry breaking ( gmsb)@xcite , the decay of the lightest neutralino ( @xmath0 ) ( next - to - lightest supersymmetric particle in these models ) into a photon and a gravitino can give rise to @xmath2 + @xmath3 ( or @xmath4 + @xmath3 ) which can have substantial event rates even after subtracting the standard model backgrounds . the signal @xmath2 + @xmath3
is in fact the most probable discovery channel of gmsb .
side by side , radiative decay of the second lightest neutralino(@xmath5 ) , in a scenario based on n = 1 supergravity ( sugra ) , into a photon and the lightest neutralino ( @xmath0 , the lightest supersymmetric particle in this class of models ) can sometimes give rise to similar final states in colliders from a pair of @xmath5 s or from a @xmath6 pair .
it requires a careful investigation to distinguish the latter type of signals from the former to establish the reality of gmsb vis - a - vis the sugra - type mode of supersymmetry breaking @xcite . in view of this , it is important to know whether there can be other new physics options which allow radiative decay of a heavier particle into a photon and another invisible particle , leading to signals of the same kind as those discussed above , and , in some cases , faking the gmsb signals which are being looked for so carefully .
it will be obvious that the closest kinematical resemblance to the gmsb signal is borne when the invisible particle is a near - massless neutral fermion , for which the immediate candidate is a neutrino .
now , radiative decay of a heavier neutral fermion into a neutrino becomes a distinct possiblity in r - parity violating susy @xcite where the lightest neutralino can decay in this vein at the one - loop level .
this motivates one to undertake an exhaustive study of the radiative decay @xmath7 in an r - parity violating scenario , to see whether in some region of the parameter space this kind of a decay can have an appreciable branching ratio in spite of the allowed tree - level decay channels of the lightest neutralino . in this paper , we have attempted such a study , taking into account r - parity violation through both bilinear and trilinear terms in the superpotential .
it should be mentioned that @xmath7 has been calculated in some earlier works @xcite where either the composition of the lightest neutralino has been confined to certain limits , or some specific regions of the susy parameter space have been adhered to .
our purpose , on the other hand , is to make a detailed scan of the parameter space , and the results presented by us highlight those particular regions where the branching ratio can be of the largest magnitude .
this , we feel , is necessary to establish the authenticity of , for example , gmsb signals during collider searches .
we also go beyond the assumption of gaugino mass unification at the scale of grand unified theories ( gut ) and present some results with the @xmath8 and @xmath9 gaugino masses treated as independent parameters , to probe whether radiative neutralino decays with enhanced rates ensue upon freeing the susy standard model from a gut embedding . and lastly , we take into account the fact that r - parity violating susy is being invoked in recent times to explain the generation of neutrino masses @xcite in consonance with the superkamiokande data on atmospheric neutrinos @xcite .
thus a section of our results pertains to that particular region of the parameter space where the indicated hierarchy of neutrino masses , together with large angle mixing between the second and the third generations , is reproduced .
the paper is organised as follows . in section 2
, we describe the basic framework of the r - parity violating model used in our calculations . in section 3 ,
some details of the loop calculation are presented .
section 4 contains our numerical results for various scenarios as well as for different choices of the parameters .
we summarise and conclude in section 5 .
detailed forms of the loop integrals are outlined in the appendix .
the fact that all the particles in the standard model carrying baryon ( b ) and lepton ( l ) number are fermions , with fixed gauge multiplet stuctures , implies that b and/or l can not be violated by one unit .
thus one can not write down renormalizable terms in the lagrangian which violate b or l or both .
this is no longer true in the case of supersymmetric theories where the particle spectrum is now doubled and baryon number and lepton number are assigned to the supermultiplets .
one can now have scalar particles containing b or l , and it is possible to have terms with @xmath10 or @xmath11 .
however , the scenario has to be made consistent with the non - observation of proton decay which requires the simultaneous violation of b and l. this is ensured in an over - restrictive manner in the minimal supersymmetric standard model ( mssm ) , by imposing a discrete multiplicative symmetry called r - parity defined as @xmath12 , which equals + 1 for standard model particles and -1 for the superpartners .
an immediate consequence of r - parity conservation is that the lightest supersymmetric particle ( lsp ) is stable . the conservation of r - parity , however , is not prompted by any strong theoretical reason , and theories where r is violated through nonconservation of _ either _ b _ or _ l are perfectly consistent with stability of the proton .
such scenarios can be studied by genaralising the mssm superpotential to the following form : @xmath13 with @xmath14 and @xmath15 in this paper we shall assume that the b - violating piece @xmath16 is not there . also , we discuss the l - violating terms in two separate categories for the convenience of analysis , considering , in turn , @xmath17 with either the bilinear ( @xmath18 ) @xcite or the trilinear ( @xmath19- and @xmath20 ) @xcite terms existing in the superpotential at a time .
@xmath21 we simplify the analysis here by assuming a bilinear r - parity violating term involving only the third leptonic generation .
all the imporatnt features of such a scenario can be seen within simplified framework . at first sight
it appears that the @xmath22-term can be rotated away from the superpotential by properly redefining the @xmath23 and @xmath24 fields , so that @xmath25 @xmath26 now as a result of this rotation the superpotential takes the form : @xmath27 & - & \frac { h_{ij}^d \epsilon } { \mu ' } { \hat q}_i { { \hat l}'}_3 { \hat d}_j^c - \frac { h_{ij}^l \epsilon } { \mu ' } { \hat l}_i { { \hat l}'}_3 { \hat e}_j^c + \frac { h_{ij}^l \mu } { \mu ' } { \hat l}_i { { \hat h}'}_1 { \hat e}_j^c \end{aligned}\ ] ] which means that the @xmath19- and @xmath20 type terms are generated in general and in the last two terms @xmath28 . but more importantly
, it should be noted that the scalar potential has terms of the following forms in the original basis : @xmath29 where we have written down the f - term and soft breaking terms relevant for our purpose here .
it is evident that such terms will generate , in general , a vacuum expectation value ( vev ) @xmath30 for the sneutrino .
the basis rotation regenerates such terms , and therefore the sneutrino vev is in general non - vanishing as the special consequence of bilinear r - parity violation whenever soft susy breaking is there @xcite .
thus there are two ways of parametrizing a bilinear r - parity violating scenario : ( i ) with @xmath22-term present in the superpotential together with a vev of the sneutrino , and ( ii ) with @xmath22-term rotated away from the superpotential but the vev of the sneutrino ( in the rotated basis ) embodying the l - violating effects that the former would imply in , say , the neutralino and chargino mass matrices .
of course , the parameter @xmath22 takes refuge in this case in the scalar potential together with the soft breaking parameter @xmath31 .
@xmath31 can be eliminated by using the conditions for electroweak symmetry breaking , if @xmath22 and @xmath30 are used as independent variables .
our calculations here are done in a basis where both @xmath22 and @xmath30 ( vev of the sneutrino ) are present in the superpotential .
now , the presence of @xmath22 and @xmath30 will induce a kind of mixing in the fermionic as well as in the scalar sector of the theory , which is typical of bilinear r - parity violation . in the fermionic sector neutralinos
will mix with the tau - neutrino and the charginos , with the tau lepton .
consequently , the neutralino mass matrix takes the following form : @xmath32 where the successive rows and columns correspond to ( @xmath33 ) . here
@xmath34 @xmath35 and @xmath36 are the @xmath37 and @xmath38 gaugino mass parameters respectively , @xmath39 , the higgsino mass parameter , and @xmath40 evidently , this will result in the generation of ( majorana ) masses for the third generation neutrino at tree - level through a see - saw type mechanism @xcite .
in such a situation , the @xmath41 chargino mass matrix is @xmath42 where @xmath43 , @xmath44 and @xmath45 , @xmath35 being the @xmath46 gaugino mass parameter .
here we have assigned @xmath47 along the rows and @xmath48 along the columns . similarly , in the scalar sector
neutral and charged scalar mass matrices are enlarged as a result of the mixing between charged sleptons and charged higgs and sneutrino with the neutral higgs . in the neutral scalar sector the scalar mass - squared matrix in the basis @xmath49 takes the form : @xmath50 in a similar way , the neutral pseudoscalar mass - squared matrix becomes @xmath51 in the basis @xmath52 and takes the form : @xmath53 in the charged scalar sector the charged scalar mass - squared matrix becomes @xmath54 in the basis @xmath55 and is given by @xmath56 with @xmath57 @xmath58 @xmath59 @xmath60 @xmath61 @xmath62 @xmath63 in addition to constraints coming from electroweak symmetry breaking , the requirement of electric charge conservation subjects the parameters in the potential to appropriate conditions in any given scenario .
in the calculations here , we make the following choice of independent parameters : @xmath64 .
we also take the ramaining scalar mass parameters to be as follows : @xmath65 and @xmath66 where , @xmath67 is the ratio of two higgs vacuum expectation values , @xmath68 is the usual b - term corresponding to the @xmath39-term , a s are the trilinear soft breaking terms .
the bilinear r - parity violating parameters @xmath22 and @xmath30 can be constrained from different experimental considerations such as the tau - mass measured within the existing errors , and , most importantly , from the constraints on neutrino masses .
the latter can be subjected to current laboratory bounds , but the restriction becomes more stringent on using the superkamiokande results on atmospheric neutrinos . in this light , the values of @xmath22 and @xmath30 have to be constrained in such a way that the sneutrino vev is less than a few hundred kev s in the basis where @xmath22 is rotated away . with @xmath69 have to be included in order to explain @xmath70 oscillation .
this requires a straightforward extension of the formalism described above . for more details ,
the reader is referred to ref . @xcite . ] the above kind of mixing in the chargino , neutralino and scalar sectors results in physical states which are superpositions of neutral(charged ) leptons and neutralinos(charginos ) on one hand , and higgses and sleptons(sneutrinos ) on the other .
this implies that the yukawa and gaugino coupling terms , written in terms of the physical states , will give rise to _ all _ the interactions of the @xmath19 and @xmath20-types .
in addition , the fact that gaugino couplings play a role here makes it possible to free some of the terms from constraints due to gauge invariance of the superpotential .
one such interaction term has particular importance in our calculation , namely one involving a top quark , a stop and a state which is dominantly a neutrino .
this is clearly a result of neutrino - neutralino mixing and the quark - squark neutralino interaction in the lagrangian . for more details the reader
is referred to @xcite . also , a scenario like this allows one to have tau - neutralino - w and neutrino - neutralino - z couplings .
this results in additional decay modes of a neutralino which are characteristic of bilinear r - parity violation @xcite .
@xmath71 in such a case the requirement of gauge invariance implies that @xmath72s are antisymmetric in the first two indices due to @xmath8 invariance but @xmath73s are not subjected to such constraints .
thus we have nine @xmath19 and twenty seven @xmath20-type couplings other than the mssm parameter space .
these trilinear couplings can also be constrained in various ways like , for example , lepton universality violation @xcite , neutrinoless double beta decay @xcite , majorana mass of neutrino @xcite , flavor changing neutral current processes @xcite etc .
their implications in various high energy collider experiments such as lep @xcite , hera @xcite or the fermilab tevatron @xcite have also been widely explored .
in addition , they can be responsible for neutrino masses generated at the one - loop level @xcite .
before we end this section , it may be remarked that although constraints on the r - violating parameters are frequently talked about , in practice these constraints will always have to stipulate a given set of values of the r - conserving ( mssm ) parameters .
the one - loop calculation of the decay @xmath74 has been performed in the nonlinear r - gauge . a similar calculation for the radiative decay of heavier neutralinos into lighter ones
can be found in ref .
@xcite . in this gauge , choice of the gauge fixing term modifies certain types of vertices compared to , say , the t hooft - feynman gauge .
for example , the @xmath75 vertex ( where @xmath76 is the charged goldstone boson ) is abesnt in the non - linear r - gauge , whereas the @xmath77 vertex gets correspondingly modified .
also a set of diagrams involving a @xmath78 transition ( with an off - shell z ) , which potentially contributes to the process under investigation , gives zero contribution when one sums over all the loops .
the diagrams which ultimately contribute in the bilinear r - parity violating scenario are the triangle diagrams shown in fig .
1 . a subset of these will be present when one considers only trilinear lepton number violating terms in the superpotential , the relevant diagrams being ( a ) and ( b ) involving only the down - type quark / squark or lepton / sleptons in the loop .
this is actually a consequence of @xmath8 invariance of the superpotential and can be seen explicitly if one expands the @xmath19- and @xmath20-type terms into their component forms .
the appearance of additional diagrams in the bilinear r - parity violating case can be attributed to mixing in the scalar and lepton - chargino - neutralino sectors , once electroweak symmetry breaking takes place ( see section 2 ) .
in fact , as has already been mentioned , one of the main reasons for undertaking this study with bilinear r - parity violation is the presence of the additional diagrams involving the top quark and the w - boson , which are potential sources of enhancement of the decay amplitude .
since the photon is on - shell , @xmath79 gauge invariance demands that the transition amplitude be proportional to @xmath80 thus one expects the following form for the matrix element : @xmath81 where @xmath82 is the mass of @xmath0 , @xmath83 and @xmath84 are the signs of the mass eigenvalues of @xmath85 and the neutrino respectively .
thus , depending on @xmath83 and @xmath84 , the effective @xmath86 interaction is either proportional to @xmath87 or @xmath88 .
the radiative decay width of @xmath0 is then given by @xmath89 where we have neglected the mass of the neutrino and @xmath90 , containing details of the loop integrals , has the dimension of inverse mass - squared . in order to calculate @xmath91
one has to evaluate the triangle graphs . for a given set of internal particles
there are two diagrams which differ from each other by the direction of charge flow in the loop . in the figures we have shown only one set of diagrams .
one must be careful about the diagrams involving the reverse flow of charge in the loop because in those cases one encounters vertices with clashing arrows .
the integrals which appear during the loop calculations are regularized using dimensional regularization .
since there is no tree - level coupling of the type @xmath86 the divergences must cancel among each gauge invariant subset of diagrams .
for example , the diagrams ( g ) and ( h ) and their partner graphs where the direction of charge flow is reversed form a gauge invariant subset and it can be shown that the infinities cancel within this set of diagrams .
similarly , one gets a finite result out of the loop diagrams involving fermion / sfermion loops , i.e. , the diagrams @xmath92 and their partner graphs .
this is true , pairwise , for the loops @xmath93 and @xmath94 involving charged higgs boson and charginos and the charged goldstone bosons and the charginos .
the decay amplitude shown in eqn.(13 ) can be decomposed in the following way ( with only the appropriate contributions retained , as stated earlier , with trilinear r - violating terms ) ; @xmath95 the matrix element due to the four diagrams ( a ) , ( b ) and their partner graphs is given by @xmath96 & { \times } & \sum_{f } e_f c_f \{(a_lb_r - a_rb_l ) [ \eta_1 { m_{\tilde \chi^0_1 } } ( i^1 - i^3 ) ] + m_f ( a_lb_l - a_rb_r)i^2\}\end{aligned}\ ] ] where @xmath97 is the charge of the fermion @xmath98 in units of e @xmath99 , and @xmath100 is the color factor for the particles in the loop .
one can use this result to include contributions from all possible fermion / scalar loops .
thus contributions from charged higgs and goldstone bosons are also present in the sum over @xmath98 .
we have also explicitly used the sum over the three possible chargino states ( @xmath101 ) in the bilinear r - parity violating scenario .
one must be careful with the masses and couplings as well as about the electric charges of the fermions in the loop .
the integrals @xmath102 are expressible in terms of the general expressions for one - loop three - point functions @xcite , in forms that are presented in the appendix , where we also give the expressions for the combinations @xmath103 and @xmath104 for all the possible cases . in the case of stop and sbottom
we have included the effect of left - right mixing .
also , the calculation includes possible absorptive parts of the loop integrals , which can be present in particular when the decaying neutralino is heavier than the @xmath105 . in a similar way ,
the contributions to the matrix element from diagrams ( g ) and ( h ) and their partner graphs are given by @xmath106 & { \times } & \sum_{k } \{(a_lb_l - a_rb_r ) [ \eta_1 { m_{\tilde \chi^0_1 } } ( i^1 - j^2 -i^3 ) ] + 2 m_k ( a_lb_r - a_rb_l)j^2\ } \end{aligned}\ ] ] in both the above expressions , the quantities @xmath107 , @xmath108 , @xmath109 and @xmath110 are all assumed to be real .
the loops involving the top quark ( fig.1(a ) and fig.1(b ) ) , which are present only in the case of bilinear r - parity violation , always seem to have a rather important effect . mixing . ] matching contributions also come over a large area of the parameter space from loops involving the w - boson ( and the tau and/or the lighter chargino , depending on the relevant mixing angles ) in the propagator ( diagrams ( g ) and ( h ) ) . the very presence of these diagrams causes the rates for radiative decays to be larger in cases with bilinear r - violating effects than in those with only trilinears , except in those where there is destructive interference between the two types of graphs .
the calculation of the branching ratio for @xmath111 involves the computation of decay widths of the lightest neutralino in all the relevant tree - level two - and three - body decay modes .
the different types of mixing that have been discussed in section 2 for bilinear r - parity violation open up the following two - body decay modes , as and when kinematically allowed : @xmath112 where h is the lightest higgs boson .
in addition , there are three body tree - level decay channels with both bilinears and trilinears . while the channels @xmath113 are available in both the cases @xcite , additional decays such as @xmath114 and @xmath115 can take place in the former , and can be important when two - body decays are not kinematically allowed . in all the numerical results presented here ,
we have used @xmath116 .
our finding is that the branching ratio of the radiative decay is rather insensitive to the values of these parameters .
let us first concentrate on bilinear r - parity violating scenario .
we will consider two different cases , namely : ( i ) when the supersymmetry breaking gaugino mass parameters are unified at the grand unification scale , and ( ii ) when @xmath8 and @xmath9 gaugino mass parameters denoted by @xmath117 and @xmath118 respectively are treated as unrelated .
due to reasons explained below , significant contributions to the radiative decay can come only when bilinear r - parity violating effects are present .
there again , the relevant parameters @xmath22 and the sneutrino vev(s ) can be constrained under two different considerations .
if the neutrino - neutralino mixing process in such a scenario is the main mechanism for the generation of ( majorana ) neutrino masses for explaining the superkamiokande ( sk ) results on atmospheric muon neutrinos , then the parameters have to be constrained in such a way that the quantity @xmath119 is less than about 100 kev _ in a basis where the @xmath22-parameters are rotated away from the superpotential_. the numerical results modulo such constraints are denoted by `` with sk '' in the corresponding figure captions , and the results correspond to both the second - and third - generation neutrinos in the final state of the radiative decays . on the other hand , if one ignores the sk constraints , then , with just the laboratory bounds on the tau neutrino mass , the values of @xmath22 and @xmath30 ( i.e. the r - violating parameters in the third generation ) can be as large as on the order of gev s . we have also presented such results , considering only the third ( tau ) neutrino to be there as the decay product .
figures 2 - 9 ( 10 ) show the results of a scan of the parameter space carried out in the different scenarios mentioned above , calculated with bilinear ( trilinear ) r - parity violation .
the first thing that we note is that in all the results presented , a specific set of masses for the squarks and sleptons are assumed .
the branching ratios of the radiative decay , however , are rather insensitive to their variation .
this is because if , for example , we reduce the squark masses , the effect of a lighter stop will increase the decay width for the radiative decay , but the corresponding tree level decays will also undergo a boost via diagrams mediated by squarks .
similarly , a comparison of figures 2 - 4 with 5 - 7 and 8 with 9 reveals that imposing the sk constraints does not cause any appreciable reduction to the probability of obtaining branching ratios on the higher side for similar combinations of mssm parameters .
all that it does is to scale the overall coefficients instrumental in both the loop - and tree - level decays .
this results in branching ratios of similar magnitudes , but is manifested in larger decay lengths for the neutralino when r - parity violating parameters are restricted to yield @xmath120 .
as has already been mentioned , the major contributions to the radiative decay come from _ ( a ) _ the top - stop loop with bilinear r - pariy violation , as well as from loops involving the w - boson and the tau or the lighter chargino .
however , this leads in some regions of the parameter space to the interesting possibility of their cancelling each other .
when such a cancellation is very severe for some specific neutralino mass , the branching ratio is seen to undergo a sharp dip at that point , as seen in fig.3 and fig.8 . in such cases ,
the effective contributions at those points hardly contain any input that is special of the bilinear terms .
on the other hand , in fig.2 ( and partially in fig.3 ) one encounters a situation where the contribution is mainly from the w - loops but the latter undergoes a destructive interference among the various component terms , causing the overall branching ratio to fall at a particular region .
the numerical results show a sensitivity to the mssm parameters @xmath39 and , to a somewhat lesser extent , on @xmath121 .
it is also clear from the graphs that the radiative decay tends to remain suppressed for small values of @xmath122 which is treated here as a free parameter .
it gradually rises with @xmath122 , and then almost saturates , showing a slight fall for @xmath122 approaching a tev .
the loops have the largest contributions for @xmath123 . by and large ,
the corresponding regions of the parameter space have the lightest neutralino almost entirely dominated by the bino state .
when the condition of gagino mass unification is relaxed , the mass parameters @xmath118 and @xmath117 , corresponding to the @xmath9 and @xmath8 gauginos can be unrelated @xcite .
most of the features of the unified scenario , including the possibilities of having branching ratios close to 10 per cent , are seen here also .
however , it entails the additional possibility of having a destructive interference between the top - and w - induced diagrams in cases where the branching ratio is otherwise on the higher side , causing the latter to fall sharply by about 5 orders of magnitudes for a particular value of the neutralino mass , with all other parameters at same values .
this is seen when @xmath117/@xmath118 is smaller than what it would have been with the constraint of unification , and the dip is found to occur when the two masses are quite close to each other .
figure 10 is a sample showing the order of magnitude of the radiative decay with only trilinear r - violating couplings present in the theory .
the values of the @xmath19- and @xmath20-type coupling constants have been used consistently with the current limits @xcite .
we have already seen that the two potentially most important classes of diagrams , namely , those mediated by the top and the w , are absent in such a case .
therefore , it is hardly surprising that the branching ratios can not rise above @xmath124 - @xmath125 over a large part of the parameter space .
the range in which the branching ratio tends to lie in such a case matches , as one would expect , with the one to which it falls when a cancellation of the large contributions takes place in the scenario with bilinears . in any case
, the radiative branching ratio turns out to be too small to be of any observable significance where r - parity is violated only through trilinear terms . with the maximum value of the branching ratio for the radiative decay being between 5 and 10 per cent , two - photon signals from such decays will fake signals like those of gmsb in only a rather small region of the parameter space .
such a thing might happen when the lighest neutralino in gmsb is close to the kinematic limit of production .
however , the fact that only the bilinear r - violating terms can boost the branching ratio to the level of close to 10 per cent has rather interesting implications in terms of observing distinctive signals of the latter .
for example , pair - produced neutralinos at lep energies ( assuming an intergated luminosity of 500 @xmath126 ) can give rise to about 40 events where there is a radiative decay on one side , leading to signals of the type @xmath127 .
there can be many more of such events in a high - energy electron - positron collider with an integrated luminosity of about 50 @xmath128 . with a proper event selection strategy
, radiative neutralino decay can thus be an interesting signature of r - parity violation with bilinear terms .
we have performed a detailed calculation of the branching ratios for the radiative decay @xmath129 for the lightest neutralino in r - parity violating susy models .
it is seen that the branching ratio can have a maximum value of about 5 - 10% when r - parity violation has its origin in the so - called ` bilinear ' terms in the superpotential , and is insignificantly small for that induced by ` trilinear ' terms .
this , we point out , can be an interesting way of obtaining characteristic signals for the former type of scenario .
our conclusion is that the chances of such radiative decays faking the two - photon signals for gauge - mediated susy breaking are small , except where the lightest neutralino in the latter is close to the kinematic limit of production .
it is also seen that the accessible region corresponding to the mssm parameter space does not change appreciably upon subjecting the theory to constraints from neutrino mass patterns required by the superkamiokande data on atmospheric neutrinos .
expressions for the mixing parameters @xmath107 , @xmath109 , @xmath108 , @xmath110 for @xmath130-loops in the bilinear r - parity violating scenario : @xmath131 @xmath132 @xmath133 @xmath134 where we have j = 2 and i = 1 .
similarly for charged scalar loops we have : \(iii ) degenerate sfermion loops : for up - type squarks we can use the same expression as that for the stop loop without the mixing angle corresponding to @xmath140 - @xmath164 mixing .
similarly for down - type squark or charged sleptons , one can use the expressions corresponding to the sbottom loop modulo the mixing factor . also the charges and the color factors should be properly included . in the case of charged sleptons
one should use the appropriate masses and charge .
one must remember that in the case of trilinear r - parity violating scenario we have diagonalised a @xmath54 neutralino mass matrix .
@xmath182 is analogous to @xmath183 but the denominator in eqn.[a.24 ] will be changed to @xmath184 the integrals which appear in the calculations of @xmath185 are identical in form to those mentioned above but one should replace @xmath186 with @xmath187 and @xmath188 with @xmath189 . here
@xmath186 and @xmath188 are the masses of the scalar and the fermion appeared in the loop respectively and @xmath189 is the mass of the chargino .
see , for example , cdf collaboration , f. abe _
et al . _ ,
lett . * 81 * , 1791(1998 ) ; phys .
rev . d*59 * , 092002(1999 ) ; d@xmath190 collaboration , s. abachi _
et al . _ ,
. lett . * 78 * , 2070(1997 ) ; d@xmath190 collaboration , b. abbott et al , phys .
lett . * 80 * , 442(1998 ) ; l3 collaboration , m. acciarri _ et al . _ ,
b * 444 * , 503(1998 ) ; opal collaboration , g. abbiendi _
et al . _ , hep - ex/9810021 .
k. grassie and p.n .
pandita , phys .
d*30 * , 22(1984 ) ; d. stump , m. wiest , and c .- p .
yuan , phys .
rev . d*54 * , 1936(1996 ) ; s. dimopoulos , m. dine , s. raby , and s. thomas , phys . rev .
* 76 * , 3494(1996 ) ; s. dimopoulos , s. thomas , and j.d . wells , phys .
d*54 * , 3283 ( 1996 ) ; s. ambrosanio , g.l .
kane , g.d .
kribs , s.p .
martin , and s. mrenna , phys .
lett * 76 * , 3498 ( 1996 ) ; phys . rev . d*54 * , 5395(1996 ) ; a. ghosal , a. kundu , and b. mukhopadhyaya , phys . rev . d*56 * , 504(1997 ) ; a. datta , a. datta , and s. raychaudhuri , eur .
j. * c1 * , 375(1998 ) .
m. dine , w. fischler , and m. srednicki , nucl .
* b189 * , 575(1981 ) ; s. dimopoulos and s. raby , _ ibid .
_ * b192 * , 353(1981 ) ; g. dvali and m. shifman , phys . lett .
b * 399 * , 60(1997 ) ; l. randall , nucl . phys .
* b495 * , 37(1997 ) e. poppitz and s. trivedi , phys .
d*55 * , 5508 ( 1997 ) ; t. han and r. hempfling , phys .
b * 415 * , 161(1997 ) ; e.j .
chun , hep - ph/9901220 ; k. choi , k. hwang , h.b .
kim , and t. lee , hep - ph/9903283 ; m. frank , h. hamidian , k. puolamaki , hep - ph/9903283 .
aulakh and r. n. mohapatra , phys .
lett b * 119 * , 136(1982 ) ; f. zwirner , phys . lett .
b * 132 * , 103(1983 ) ; l.j . hall and m. suzuki , nucl .
* b231 * , 419(1984 ) ; g. ross and j.w.f .
valle , phys .
lett b * 151 * , 375(1985 ) ; s. dawson , nucl . phys . * b261 * , 297(1985 ) e.j .
chun , s.k .
kang , c.w .
kim , and u.w .
lee , hep - ph/9807327 ; v. bednyakov , a. faessler , and s. kovalenko , phys .
b * 442 * , 203(1998 ) ; a. s. joshipura and s.k .
vempati , hep - ph/9808232 ; o. kong , hep - ph/9808304 ; d.e .
kaplan , a.e .
nelson , hep - ph/9901254 ; a. joshipura and m. nowakowski , phys .
rev . d*51 * 2421(1995 ) ; _ ibid _ , 5271(1995 ) ; f. vissani and a.yu .
smirnov , nucl .
. * b460 * 37(1996 ) ; m. nowakowski and a. pilaftsis , nucl . phys . *
b461 * 19(1996 ) ; r. hempfling , nucl . phys .
* b478 * 3(1996 ) ; t. banks , y. grossman , e. nardi , and y. nir , phys .
rev . d*52 * 5319(1996 ) ; b. de carlos and p.l .
white , phys .
* 3424(1996 ) ; h. p. nilles and n. polonsky , nucl . phys . *
b484 * 33(1997 ) .
b. de carlos and p.l .
white , phys .
rev . d*55 * 4222(1997 ) ; e. nardi , phys .
rev . d*55 * 5772(1997 ) ; a. akeroyd , m.a .
daz , j. ferrandis , m.a .
garca - jareo , and j. w. f. valle , hep - ph/9707395 ; a. faessler , s. kovalenko and f. imkovic , phys .
d*58 * 055004(1998 ) ; m.a .
daz , j. ferrandis , j.c .
romo , and j.w.f .
valle , hep - ph/9801391 .
tai - fu feng , hep - ph/9808379 ; m.a .
daz , e. torrente - lujan , j.w.f .
valle , hep - ph/9808412 ; j. ferrandis , hep - ph/9810371 ; m. bisset , o.c.w .
kong , c. macesanu , l.h .
orr , hep - ph/9811498 ; m. hirsch and j.w.f .
valle , hep - ph/9812463 ; c - h .
chang and t - f .
feng , hep - ph/9901260 ; f. de campos _ et al .
_ , hep - ph/9903425 .
k. enqvist , a. masiero , and a. riotto , nucl . phys .
* b373 * 95(1992 ) ; f.m .
borzumati , y. grossman , e. nardi , and y. nir , phys . lett . *
b384 * 123(1996 ) ; g. bhattacharyya , susy 96 , nucl . phys . * b * ( proc .
* 52a * , 83(1997)(hep - ph/9709395 ) ; h. dreiner , hep - ph/9707435 s. dimopoulos and l.j .
hall , phys .
b * 207 * , 210(1987 ) ; r.m .
godbole , p. roy , and x. tata , nucl . phys . *
b401 * , 67(1993 ) ; j. kalinowski , r. rckl , h. spiesberger and p.m. zerwas , phys .
b * 406 * , 314(1997 ) . l. hall and m. suzuki in ref .
@xcite ; m. drees , s. pakvasa , x. tata , and t. ter veldhuis , phys .
rev . d*57 * , 5335(1998 ) ; r. adhikari and g. omanovic , phys .
rev . d*59 * , 073003(1999 ) ; s. rakshit , g. bhattarchyya , and a. raychaudhuri , phys .
d*59 * , 091701(1999 ) g. t hooft and m. veltman , nucl . phys . *
b153 * , 365(1979 ) ; g. passarino and m. veltman , nucl . phys . *
b160 * , 151(1979 ) .
the calculations have been done using the algorithm developed in b. mukhopadhyaya and a. raychaudhuri , phys .
rev . d*39 * , 280(1989 ) . | in an r - parity violating supersymmetric scenario , the lightest neutralino @xmath0 is no longer a stable particle .
we calculate the branching ratio for the decay mode @xmath1 which occurs at the one - loop level . taking into account bilinear as well as trilinear lepton number violating interactions as the sources of r - parity violation
, we make a detailed scan of the parameter space , both with and without gaugino mass unification and including the constraints on the neutrino sector from the recent superkamiokande results .
this study enables one to suggest interesting experimental signals distinguishing between the two types of r - parity breaking , and also to ascertain whether such radiative decays can give rise to collider signals of the type @xmath2 + @xmath3 from pair - produced neutralinos .
= -6pt = -1 in = 6.5 in = 9.4 in = 0.75 in 2 q^2 // mri - phy / p990307 + march , 1999 + hep - ph/9903418 * radiative decay of the lightest neutralino in an r - parity violating supersymmetric theory * + biswarup mukhopadhyaya and sourov roy + _ mehta research institute , + chhatnag road , jhusi , allahabad - 211 019 , india _
+ 1 true cm pacs nos . : 12.60.jv , 13.10.+q , 14.80.ly |
in the last few years , approximately 40 new galactic sources of very - high - energy ( vhe ) @xmath2-ray emission ( 100gev@xmath3100tev ) have been discovered using the high energy stereoscopic system ( hess ) cherenkov telescope array ( e.g. , * ? ? ?
these are an exciting new population of sources , which give new insight into non - thermal particle acceleration in galactic objects such as neutron stars , supernova remnants , and x - ray binaries . thus far ,
close to half of these sources have been established or suggested as being associated with the pulsar wind nebulae ( pwne ) of young pulsars , either through direct detection of a pwn or positional coincidence with a young pulsar which is presumed to have a pwn ( table [ tab : assoc ] ) .
clearly , pwne are now an important and well - established galactic source of vhe @xmath2-rays . since both young pulsars and their pwne can be very dim , many of the galactic hess sources without identified counterparts ( e.g. , * ? ? ?
* ) may be pwne , potentially detectable via deep radio or x - ray observations . in general ,
radio / x - ray pwne are associated with extended hess sources , presumably tev pwne , whose peak is offset by several arcminutes from the pulsar ( table [ tab : assoc ] ) . in some cases
the offset , if any , can not be measured because the position of the pulsar is not known ( see also * ? ? ?
these offsets have been explained by the hypothesis that the vhe emission is from inverse compton scattering of `` old '' electrons , which were produced during an earlier epoch in the pulsar s life , off of the ambient photon field ( e.g. , cosmic microwave background radiation , starlight , and infrared emission from dust ; see aharonian et al .
2005d , de jager & djannati - ata 2008 ) .
an alternative model has the @xmath2-rays produced by the decay of @xmath6 mesons , which are created by the interaction of accelerated hadrons with nuclei in the interstellar medium @xcite . in both cases , asymmetric crushing of the pulsar wind by the reverse shock of the expanding supernova remnant , in cases where
the remnant has expanded into an asymmetric density distribution in the ambient medium , is then responsible for the offset between the pulsar and the peak of the vhe emission @xcite .
a high pulsar proper motion may also play a role in the offset .
the vhe emission could be a key into the earlier energetic history of these young pulsars as well as completing a broadband ( radio to tev @xmath2-rays ) picture of the shocked pulsar wind , whose emission would be dominated by synchrotron - emitting electrons below @xmath7gev and inverse compton emission from the scattering of these same particles above this energy . in this letter ,
we present the discovery and subsequent timing of the `` vela - like '' pulsar ( i.e. pulsars having characteristic age @xmath8kyr and spin - down luminosity @xmath9ergs s@xmath1 ; e.g. , * ? ? ?
* ) psr j1856 + 0245 in the arecibo palfa survey of the galactic plane .
this young , energetic pulsar is positionally coincident with , and energetically capable of creating , the vhe emission observed from the hitherto unidentified tev source hess j1857 + 026 @xcite .
this association suggests that some of the other currently unidentified , extended hess sources in the galactic plane may also be related to faint radio pulsars , rather than some new source class .
an archival _ asca _ image of the area around psr j1856 + 0245and hess j1857 + 026 shows a possible x - ray counterpart , cataloged as ax j185651 + 0245 by @xcite , at the pulsar position ; this is possibly a synchrotron counterpart to the tev pwn @xmath2-ray data , @xcite identified the source egr j1856 + 0235 , which is analogous to 3eg j1856 + 0114 in the 3rd @xmath10 catalog .
psr j1856 + 0245 is well within the 95% confidence region of egr j1856 + 0235 , suggesting the pulsar and/or its pwn are also visible in the mev - gev range .
we will investigate this further in a follow - up paper . ] .
psr j1856 + 0245 was discovered in the arecibo palfa survey for pulsars and radio transients ( cordes et al .
2006 ; see also hessels 2007 ) .
palfa is using the 1.4-ghz arecibo l - band feed array ( alfa ) 7-beam receiver to survey the galactic plane at longitudes of @xmath11 and @xmath12 , out to latitudes @xmath13 .
the relatively high observing frequency and unequaled raw sensitivity of the arecibo telescope make the palfa survey sensitive to distant , faint , and scattered pulsars that were missed in previous surveys .
the larger goals , design , and observational setup of the palfa survey are presented in detail by @xcite .
we found psr j1856 + 0245 in a 268-s survey observation made on 2006 april 16 .
the pulsar was identified at a signal - to - noise ratio of 35 within a few minutes of the discovery observation itself through a `` real time '' processing pipeline , operating on data with reduced time and spectral resolution , which is automatically run on the survey data as it is being collected @xcite .
psr j1856 + 0245 has a spin period of 81ms and a large dispersion measure ( dm ) , 622@xmath14 pc .
we estimate that the flux density at 1400mhz is @xmath15mjy .
we also note that psr j1856 + 0245 shows a significant scattering tail at 1170mhz ( @xmath16ms at this frequency , where @xmath17 is the time constant of a one - sided exponential fitted to the pulse profile ) and would be difficult to detect below @xmath18mhz because the scattering time scale would be greater than the pulse period . immediately following the discovery of psr j1856 + 0245
, we began regular timing observations with arecibo and the jodrell bank observatory s 76-m lovell telescope in order to derive a precise rotational and astrometric ephemeris .
between the two telescopes , timing observations were made on 148 separate days between 2006 april 24 and 2008 april 18 . at arecibo
, observations were made using both the center pixel of the alfa receiver at 1440mhz ( identical setup to the standard palfa survey mode ) and the l - wide receiver with multiple wide - band arecibo pulsar processor ( wapp ) correlators centered at 1170 , 1370 , 1470 , and 1570mhz , each with 256 lags over 100mhz of bandwidth , sampled every 256@xmath19s .
the typical integration time was 3 minutes .
the jodrell bank observations were made at 1402/1418mhz with a @xmath20mhz incoherent filterbank system and 202@xmath19s sampling time .
the typical integration time was 20 minutes .
the resulting phase - connected timing solution for psr j1856 + 0245 is presented in table [ tab : pulsar ] and combines data from both arecibo and jodrell bank .
this solution shows that psr j1856 + 0245 is a vela - like pulsar with a characteristic age of 21kyr and a spin - down luminosity of @xmath21ergs s@xmath1 .
the solution was obtained using the tempo pulsar timing package .
the times of arrival ( toas ) can be phase - connected by fitting only for position , dm , spin period @xmath22 , period derivative @xmath23 , and period second derivative @xmath24 , though further higher - order period derivatives are required to remove all trends in the toas ( i.e. , to `` whiten '' the residuals ) .
these higher - order period derivatives , including @xmath24 , are non - determinisitc and are likely the result of timing noise , which is common in young pulsars .
timing noise affects the measured position of psr j1856 + 0245 at a level exceeding the formal tempo uncertainties by an order of magnitude . separately fitting the toas from the first and second years of timing gives positions which differ by @xmath25 .
we take this as a rough measure of the true uncertainty on the timing - derived position of psr j1856 + 0245 . the identification of psr j1856 + 0245 as a young
, energetic pulsar raises the likelihood that it powers a radio and/or x - ray pwn ( for a review , see * ? ? ?
accordingly , we have checked source catalogs and archival , multi - wavelength data for other sources at the radio timing position of the pulsar . these searches revealed that psr j1856 + 0245 is spatially coincident with the vhe @xmath2-ray source hess j1857 + 026 and the faint x - ray source ax j185651 + 0245 .
we discuss these possible associations in 3 .
psr j1856 + 0245 is spatially coincident with the unidentified vhe @xmath2-ray source hess j1857 + 026 discovered by ( * ? ? ?
* see fig .
[ fig : xray ] ) . here
we show that psr j1856 + 0245 is energetically capable of powering hess j1857 + 026 and that this association has similar characteristics to the other pulsar / vhe associations in the literature ( table [ tab : assoc ] ) .
the estimated distance to psr j1856 + 0245 , based on its dm and the ne2001 electron density model of the galaxy @xcite , is @xmath26kpc .
the uncertainty on this distance is not well - defined , but in some cases the model can be off by as much as a factor of @xmath27 . adopting the dm - distance of 9kpc
gives a large spin - down flux of @xmath28ergs s@xmath1 kpc@xmath29 , where @xmath30 is the true distance scaled to 9kpc .
@xcite find that statistically 70% of pulsars with @xmath31ergs s@xmath1 kpc@xmath29 are visible as vhe @xmath2-ray sources .
psr j1856 + 0245 is very close to this limit and may exceed it if its distance is overestimated .
hence , based on its energetics alone , it is likely to be visible as a vhe @xmath2-ray source .
hess j1857 + 026 has a photon index @xmath32 and @xmath5tev flux @xmath33ergs cm@xmath29 s@xmath1 ( about 15% of the crab s flux in this energy range ) ; these spectral parameters are similar to those measured for the other hess sources identified with pwne .
given the spin - down luminosity of psr j1856 + 0245 , this suggests an efficiency @xmath35% ( @xmath5tev ) , comparable to what is seen in other proposed associations ( table [ tab : assoc ] ) .
psr j1856 + 0245 is offset from the centroid of hess j1857 + 026 , @xmath36 , @xmath37 ( j2000 , there is a @xmath38 statistical uncertainty on this position ) , by @xmath39 ( fig .
[ fig : xray ] ) .
as discussed in 1 , this is most likely explained by asymmetric confinement of the pulsar wind .
this interpretation is supported by psr j1856 + 0245 s offset location on the side of hess j1857 + 026 that appears somewhat compressed ( i.e. , there is a steep gradient in the @xmath2-rays ) compared with the rest of the nebula .
if , however , the offset of the vhe emission is due primarily to the proper motion of psr j1856 + 0245 , then the direction and rough magnitude of this motion are predictable . based on its characteristic age and
offset from the centroid of hess j1857 + 026 , psr j1856 + 0245 s proper motion should be roughly 23mas yr@xmath1 ( transverse velocity @xmath40 km s@xmath1 ) , to the north - west , assuming that the centroid of hess j1857 + 026 marks the birthplace of the pulsar .
the velocity is very large , but not unprecedented for a pulsar @xcite . of course
, the velocity will be smaller if the distance to the source is overestimated , or if the offset is at least partially due to an asymmetrically confined pulsar wind .
detecting this proper motion via timing or interferometry would elucidate this further and may be possible in the coming years , though timing noise and the low flux of the pulsar will make this difficult .
psr j1856 + 0245 and hess j1857 + 026 are also coincident with the faint _ asca_x - ray source ax j185651 + 0245 reported by ( * ? ? ?
* see fig .
[ fig : xray ] ) .
ax j185651 + 0245 was found @xmath38 off axis in observations from 1998 april 6 ( sequence number 56003000 ) .
it was detected only in the hard band ( @xmath41kev ) of the gas imaging spectrometers ( giss ) , with a count rate of 2.6ks@xmath1 gis@xmath1 and a significance of 4.3@xmath42 .
the exposure was @xmath43ks for each of gis2 and gis3 .
psr j1856 + 0245 and ax j185651 + 0245 ( j2000 position : @xmath44 , @xmath45 ) are spatially coincident to within the @xmath46 positional uncertainty of sources in the @xcite catalog .
although the signal - to - noise ratio of the detection of ax j185651 + 0245 is modest , its exact spatial coincidence with a young pulsar of relatively high spin - down flux argues that this source is real .
most of the previously established associations of hess sources with young pulsars also have known x - ray synchrotron pwne .
+ 0245could be an x - ray pwn created by psr j1856 + 0245 . using cxcpimms , with an
assumed absorbed power law spectrum with column density @xmath47cm@xmath29 ( roughly the total galactic contribution in this direction ) and photon index @xmath48 ( typical for x - ray pwne ) , we find that ax j185651 + 0245 has an unabsorbed flux ( @xmath41kev ) of @xmath49ergs s@xmath1 cm@xmath29 .
this corresponds to an efficiency for the conversion of spin - down energy into x - rays @xmath50% ( @xmath41kev ) that falls into the observed range for vela - like pulsars @xcite .
there are two additional nearby sources detected by @xcite : ax j185721 + 0247 and axj185750 + 0240 .
it is possible that these are part of some extended x - ray emission related with hess j1857 + 026 , though deeper x - ray observations are needed to investigate this ( see below ) .
there are four short _ swift
_ observations of the region containing psr j1856 + 0245 , including two observations specifically targeting ax j185651 + 0245 .
the deepest of these , from 2007 march 13 ( observation i d 36183002 ) , is a 4.1-ks on - axis exposure with the _ swift _ x - ray telescope ( xrt ) .
this observation does not show any significant emission at the pulsar position ; within a circular extraction region of radius @xmath46 around the pulsar , the background subtracted number of counts is @xmath51 . using the aforementioned count rate of ax j185651 + 0245 , along with the same assumed spectrum , the predicted @xmath52kev count rate for xrt from cxcpimms is @xmath53counts / ks .
thus , in 4.1ks , there should have been @xmath54counts , consistent with the 2-@xmath42 upper limit derived from the data .
thus , if ax j185651 + 0245 is predominantly a point source , it was only at the limit of detectability in this observation .
if it is predominantly an extended nebula , then it would not have been detectable in such a short exposure .
clearly , future , deeper observations , like those that recently discovered a likely x - ray pwn associated with hess j1718@xmath3385/psr j1718@xmath33825 @xcite , will be needed to determine the nature of this candidate x - ray pwn .
we have been granted a @xmath55-ks _ xmm - newton _ observation of psr j1856 + 0245 and will present an analysis of those data in a follow - up paper . at least half of the known hess / pulsar associations are accompanied by radio emission classified as a pwn , a notable exception being hess j1825@xmath3137/psr b1823@xmath313 .
only a few of the extended hess sources , e.g. hess j0835@xmath3455 , hess j1813@xmath3178 , and hess j1640@xmath3465 , are known to be accompanied by a supernova remnant ( snr ) .
we have checked available radio imaging data for signs of a pwn or snr .
there is some faint , extended emission in the vicinity of psr j1856 + 0245 visible in 1.4-ghz vla galactic plane survey data ( vgps , * ? ? ?
* ) , though nothing that is clearly indicative of a pwn or snr .
the surface brightness limit from vgps is @xmath56jy sr@xmath1 at 1.4ghz .
it is certainly not uncommon for vela - like pulsars to have faint or no known radio nebula ( e.g. psr b1823@xmath313 , * ? ? ?
* ; * ? ? ?
roughly a third of the cataloged galactic snrs @xcite are fainter than the surface brightness limit achieved by the vgps .
deep , dedicated radio imaging observations of psr j1856 + 0245 are necessary to investigate this further .
higher resolution 1.4-ghz data from the magpis survey ( helfand et al .
2006 ; in the vicinity of psr j1856 + 0245 , this survey has a sensitivity of 0.2mjy / beam at an angular resolution of @xmath57 ) reveal no point source which can be definitively associated with psr j1856 + 0245 , as expected given the flux density and positional uncertainty of the pulsar .
the arecibo observatory , a facility of the naic , is operated by cornell university in a cooperative agreement with the national science foundation .
we thank karl kosack and the hess collaboration for providing the @xmath2-ray image of hess j1857 + 026 . this work was supported by nserc ( cgs - d , pdf , and discovery grants ) , the canadian space agency , the australian research council , fqrnt , the canadian institute for advanced research ,
the canada research chairs program , the mcgill university lorne trottier chair in astrophysics and cosmology , and nsf grants ast-0647820 and ast-0545837 .
lccccccccc j0534 + 220 & & b0531 + 21 & & 33 & 1 & 460 & 2 & 0.009 & 1 + j0835@xmath3455 & 26 & b0833@xmath345 & 18 & 89 & 11 & 6.9 & 0.3 & 0.008 & 2 + j0852@xmath3463 & 60 & j0855@xmath34644 & 43 & 65 & 141 & 1.1 & _ 2 _ & 3 & 3 + j1303@xmath3631 & 10 & j1301@xmath36305 & 11 & 185 & 11 & 1.7 & _ 7 _ & 5 & 4 + j1418@xmath3609 & 4 & g313.3 + 0.1 & 8 & & & & & & 5 + j1420@xmath3607 & 4 & j1420@xmath36048 & 3 & 68 & 13 & 10 & _ 6 _ & 0.4 & 5 + j1514@xmath3591 & 6 & b1509@xmath358 & 2 & 151 & 16 & 18 & _ 4 _ & 0.2 & 6 + j1616@xmath3508 & 8 & j1617@xmath35055 & 10 & 69 & 8 & 16 & _ 7 _ & 0.6 & 7 + j1640@xmath3465 & 3 & [email protected] & & & & & & & 8 + j1702@xmath3420 & 18 & j1702@xmath34128 & 35 & 182 & 55 & 0.34 & _ 5 _ & 5 & 9 + j1718@xmath3385 & 9 & j1718@xmath33825 & 8 & 75 & 90 & 1.3 & _ 4 _ & 0.4 & 10 + j1747@xmath3281 & & g0.9 + 0.1 & & & & & & & 11 + j1804@xmath3216 & 12 & b1800@xmath321 & 10 & 134 & 16 & 2.2 & _ 4 _ & 2 & 12 + j1809@xmath3193 & 32 & j1809@xmath31917 & 12 & 83 & 51 & 1.8 & _ 4 _ & 1 & 13 + j1813@xmath3178 & 2 & [email protected] & & & & & & & 14 + j1825@xmath3137 & 10 & b1823@xmath313 & 11 & 101 & 21 & 2.8 & _ 4 _ & 1 & 15 + j1833@xmath3105 & & j1833@xmath31034 & & 62 & 5 & 34 & _ 3 _ & 0.008 & 16 + j1837@xmath3069 & 7 & j1838@xmath30655 & 6 & 70 & 23 & 5.5 & 7 & 1 & 17 + j1846@xmath3029 & & j1846@xmath30258 & & 326 & 1 & 8.1 & 6 & 0.1 & 16 + j1857 + 026 & 7 & j1856 + 0245 & 8 & 81 & 21 & 4.6 & _ 9 _ & 3 & 18 + j1912 + 101 & 16 & j1913 + 1011 & 9 & 36 & 170 & 2.9 & _ 5 _ & 0.6 & 19 + right ascension @xmath60 ( j2000 ) & @xmath61 + declination @xmath62 ( j2000 ) & @xmath63 + galactic longitude @xmath64 & 36008 + galactic latitude @xmath65 & + 0058 + pulse period @xmath22 ( s ) & 0.080902591336(5 ) + period derivative @xmath66 & @xmath67 + period second derivative @xmath68 ( s@xmath1 ) & @xmath69 + dispersion measure dm ( @xmath14 pc ) & 622(2 ) + + distance @xmath70 ( kpc ) & @xmath26 + spin - down luminosity @xmath71 ( ergs s@xmath1 ) & @xmath72 + surface dipole magnetic field @xmath73 ( g ) & @xmath74 + characteristic age @xmath75 ( kyr ) & 21 + | we present the discovery of the vela - like radio pulsar j1856 + 0245 in the arecibo palfa survey .
psr j1856 + 0245 has a spin period of 81ms , a characteristic age of 21kyr , and a spin - down luminosity @xmath0ergs s@xmath1 .
it is positionally coincident with the tev @xmath2-ray source hess j1857 + 026 , which has no other known counterparts .
young , energetic pulsars create wind nebulae , and more than a dozen pulsar wind nebulae have been associated with very - high - energy ( 100gev@xmath3100tev ) @xmath2-ray sources discovered with the hess telescope . the @xmath2-ray emission seen from hess j1857 + 026 is potentially produced by a pulsar wind nebula powered by psr j1856 + 0245 ; faint x - ray emission detected by _
asca _ at the pulsar s position supports this hypothesis .
the inferred @xmath2-ray efficiency is @xmath4% ( @xmath5tev , for a distance of 9kpc ) , comparable to that observed in similar associations . |
in an attempt to unify electrodynamics and general relativity , kaluza introduced an extra , a fifth dimension @xcite and assumed all fields to be independent of the extra dimension .
klein @xcite followed this idea , however , he assumed the fifth dimension to be compactified on a circle of planck length .
the resulting theory describes @xmath0-dimensional einstein gravity plus maxwell s equations .
one of the new fields appearing in this model is the dilaton , a scalar companion of the metric tensor . in an analogue way
, this field arises in the low energy effective action of superstring theories and is associated with the classical scale invariance of these models @xcite . both string theories @xcite as well as
so - called `` brane worlds '' @xcite assume that space - time possesses more than four dimensions .
in string theories , these extra dimensions are -following the idea of klein- compactified on a scale of the planck length , while in brane worlds , which assume the standard model fields to be confined on a 3-brane , they are large and even infinite
. it should thus be interesting to study classical solutions of non - abelian gauge theories in higher dimensions .
an einstein - yang - mills model in @xmath1 dimensions was studied recently @xcite . assuming the metric and matter fields to be independent on the extra coordinates
, an effective 4-dimensional einstein - yang - mills - higs - dilaton model appears with one higgs triplet and one scalar dilaton .
this idea was taken further to @xmath2 dimensions @xcite , where @xmath3 higgs triplets and @xmath3 dilatons appear .
in contrast to one extra dimension , however , it appeared that in two or more extra dimensions , constraints result from the off - diagonal terms of the energy - momentum tensor .
this leads to the fact that only solutions with either only one non - zero higgs field or with all higgs fields constant are allowed . in this paper
, we study an einstein - yang - mills model in @xmath4 dimensions and we introduce an off - diagonal term in the metric associated with the extra dimensions in order to be able to obtain non - trivial solutions .
we introduce the model in section ii and give the ansatz and equations of motion in section iii . in section iv , we give the 4-dimensional effective action from which the equations of motion can equally well be derived . in section
v , we present our numerical results and finally , in section vi we give our conclusions .
the einstein - yang - mills lagrangian in @xmath5 dimensions is given by : @xmath6 with the su(2 ) yang - mills field strengths @xmath7 , the gauge index @xmath8 and the space - time index @xmath9 .
@xmath10 and @xmath11 denote respectively the @xmath12-dimensional newton s constant and the coupling constant of the gauge field theory .
@xmath10 is related to the planck mass @xmath13 by @xmath14 and @xmath15 has the dimension of @xmath16 ^ 2 $ ] .
in this paper , we will construct solutions with off - diagonal components of the metric tensor .
the ansatz for the metric then reads : @xmath17 + 2 e^{\xi_1+\xi_2 } \sinh\left(\frac{j}{2}\right ) dx^4 dx^5 \ , \end{aligned}\ ] ] where the functions @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 depend on the variable @xmath23 only and @xmath24 .
the determinant of the metric is then given by : @xmath25 the @xmath26 , @xmath27 play the role of two scalar dilatons .
the ansatz for the gauge field reads : @xmath28 note that the @xmath29 , @xmath30 play the role of higgs fields
. the spherically symmetric ansatz is given by @xcite : @xmath31 @xmath32 @xmath33 where @xmath34 is a mass scale determining the vacuum expectation values of the higgs fields .
using the metric ( [ metric ] ) , the matter lagrangian @xmath35 reads : @xmath36 \nonumber \\ & - & \frac{1}{2}\sinh\left(\frac{j}{2}\right ) f_{\mu 5}^a f_4^{\mu , a } -\frac{1}{2}e^{-2\xi_1 - 2\xi_2}f_{45}^af_{4}^{5,a } \ \ , \ \
\mu , \nu=0,1,2,3 \ .\end{aligned}\ ] ] the last term in ( [ reformulation ] ) vanishes since @xmath37 this is due to the fact that the fields @xmath38 , @xmath39 are assumed to be parallel . with @xmath40
the matter lagrangian @xmath35 reads : @xmath41 inserting the anstze ( [ ansatz1])-([higgsansatz ] ) and using @xmath42 this reads : @xmath43 + a_{12 } + 2c_{12}\ ] ] with the abbreviations : @xmath44 the non - vanishing components of the energy - momentum tensor are given by : @xmath45 \\ & t_1 ^ 1&=e^{\xi_1+\xi_2}\left [ b-\frac { d}{2}+\frac { a_1}{2}+\frac { a_2}{2}- c_1- c_2 + a_{12 } - 2 c_{12 } \right]\\ & t_2 ^ 2=t_3 ^ 3&=-e^{\xi_1+\xi_2}\left[\frac { a_1}{2}+\frac { a_2}{2}-\frac { d}{2 } - a_{12 } \right]\\ & t_4 ^ 4&=-e^{\xi_1+\xi_2}\left [ b+\frac { d}{2}-\frac { a_1}{2}+\frac { a_2}{2}- c_1 + c_2 \right]\\ & t_5 ^ 5&=-e^{\xi_1+\xi_2}\left [ b+\frac { d}{2}+\frac { a_1}{2}-\frac{a_2}{2}+ c_1- c_2 \right]\\ & t_4 ^ 5&=-e^{2\xi_1}\left[\tanh\left(\frac{j}{2}\right ) ( a_1 + 2 c_1 ) - \coth\left(\frac{j}{2}\right ) ( a_{12 } + 2 c_{12 } ) \right]\\ &
t_5 ^ 4&=-e^{2\xi_2}\left[\tanh\left(\frac{j}{2}\right ) ( a_2 + 2 c_2 ) - \coth\left(\frac{j}{2}\right ) ( a_{12 } + 2 c_{12 } ) \right]\end{aligned}\ ] ] the presence of a off - diagonal term in the metric renders the components of the einstein tensor very lengthly and we do not give them explicitely here .
we note that these off - diagonal terms have to be included for codimension larger than one to avoid solutions with trivial higgs fields .
this is a new feature with respect to @xcite .
five independent einstein equations can be obtained for the five metric functions parametrizing the metric ( with @xmath46 ) : @xmath47 \nonumber \\ & + & \frac{1}{32}x^2n\left[(j')^2 + 10 ( \xi_1')^2 + 10 ( \xi_2')^2 + 12\xi_1'\xi_2 ' \right ] \nonumber \\ & + & \frac{1}{16 } \cosh\left(\frac{j}{2}\right ) x^2 n \left(\xi_1'- \xi_2'\right)^2 \ \ , \end{aligned}\ ] ] @xmath48 + \frac{1}{16}ax\left[(j')^2 + 10(\xi_1')^2 \right .
\nonumber \\ & + & \left .
10(\xi_2')^2 + 12\xi_1'\xi_2 ' + 2 \cosh\left(\frac{j}{2}\right ) x^2 n ( \xi_1'- \xi_2')^2\right ] \ \ , \end{aligned}\ ] ] @xmath49 \ \ , \end{aligned}\ ] ] @xmath50 \ , \end{aligned}\ ] ] where we use the abbreviation @xmath51 and the prime denotes the derivative with respect to @xmath52 .
note that the equation for @xmath22 can be obtained for ( [ einstein 3 6d ] ) by exchanging @xmath53 .
finally the variation of the action ( [ action ] ) with respect to the matter fields leads to three differential equations for the functions @xmath54 , @xmath55 and @xmath56 : @xmath57 \nonumber \\ & - & 2 a h_1 h_2 k e^{-\xi_1-\xi_2 } \sinh\left(\frac{j}{2}\right ) \ , \end{aligned}\ ] ] @xmath58 where @xmath59 and @xmath60 .
we will study globally regular , asymptotically flat solutions of the system above .
this implies the following boundary conditions : @xmath61 at the origin and @xmath62 at infinity .
the above equations can be obtained equally from the following 4-dimensional effective action : @xmath63\nonumber\\ & & -\frac{1}{2}(\partial_\mu\psi_1)(\partial^\mu\psi_2 ) -\frac{1}{12}\cosh{(j)}~[~(\partial_\mu\psi_1)(\partial^\mu\psi_1 ) + ( \partial_\mu\psi_2)(\partial^\mu\psi_2)\nonumber\\ & & -2(\partial_\mu\psi_1)(\partial^\mu\psi_2)~]-\frac{1}{32}(\partial_\mu j)(\partial^\mu j ) \end{aligned}\ ] ] provided the following identification is done @xmath64 note that @xmath65 is a new coupling constant ( the `` dilaton '' coupling ) appearing in the effective action .
the function @xmath20 appears as a new dynamical scalar field in the effective action .
the system of non - linear differential equations can only be solved numerically .
we have solved this system for numerous values of @xmath66 . for the numerical analysis
we assume @xmath67 , but we believe that the pattern of solutions remains qualitatively equal for @xmath68 . in the limit @xmath69 the metric is flat ( minkowski metric ) ( implying @xmath70 ) and the solution of the equations is the t hooft - polyakov magnetic monopole @xcite . due to the presence of two -linearly superposed- higgs fields , the flat solution has mass @xmath71 , which is confirmed by our numerical results . for @xmath72 the solution is progressively deformed by gravity , the fields @xmath73 , @xmath74 , @xmath75 , @xmath76 become non - trivial .
in particular , @xmath76 develops a local minimum at some finite radius @xmath77 , @xmath78 represents the minimum of the function @xmath75 , the dilaton field @xmath79 is negative and monotonically increasing , while the function @xmath73 is positive and decreases monotonically from @xmath80 to @xmath81 .
the profiles of the different functions is presented in fig.[fi45f1 ] for two different values of @xmath66 .
this figure shows that @xmath82 is larger than @xmath83 .
the values @xmath84 and @xmath85 are plotted as functions of @xmath66 in fig .
[ fi45f2 ] and fig .
[ fi45f4 ] , respectively .
the branch which arises from the gravitational deformation of the flat monopole is indexed by the label `` 1 '' in the figures . in fig.[fi45f2 ] we also plot the adm mass of the solution , which decreases with increasing @xmath66 .
the figures further demonstrate that the branch `` 1 '' of solutions does not exist for arbitrary large values of @xmath66 .
rather , the solutions exist only for @xmath86 with @xmath87 .
in addition , another branch of solutions exists for @xmath88 .
more precisely , we have managed to construct a second branch of solution ( labelled `` 2 '' in the figures ) for @xmath89 ( with @xmath90 ) .
again , starting from @xmath91 and increasing @xmath66 , our numerical analysis strongly suggests the existence of a third branch labelled `` 3 '' .
note that the adm mass of the solutions of branch `` 3 '' is very close to that of the solutions of branch `` 2 '' .
this is why the two branches can not be distinguished on the plot as far as their adm mass is concerned . continuing this construction ,
there probably exist a large ( possibly infinite ) number of branches on smaller and smaller intervals concentrated around the value @xmath92 .
this pattern is very reminiscent to the one occuring in the case with @xmath93 first discussed by volkov @xcite .
however the behaviour of the mass of the solutions in the @xmath93 case and in the @xmath5 cases is considerably different . indeed , for @xmath93 ,
the mass of the solutions on the branches `` 2'',``3 '' is higher than the mass of the solution on the branch `` 1 '' ; in the present case , it is the contrary .
said in other words , for @xmath94 the solution with the lowest mass is the one of branch `` 2 '' . at first glance
this result may appear paradoxal , since it is known that the monopole is topologically stable .
however , we believe that this peculiar behaviour of the mass is deeply connected to the direct coupling between the two higgs fields in the effective action . roughly speaking the two higgs fields
are coupled with a factor @xmath95 ( remember that this term is absent in absence of gravity @xmath96 ) . for @xmath97
it turns out that the three contributions due to the terms quadratic in the derivative of the higgs fields have a tendency to compensate . for
the solutions of the second branch , the dilatons @xmath98 and the function @xmath73 deviate stronger from their flat space - time values than on the branch `` 1 '' .
specifically , @xmath73 becomes large enough to dimnish considerably the contribution of the higgs field to the mass . as a consequence , the mass becomes smaller than the mass on the first branch . analyzing this result from the point of view of catastrophe theory it it tempting to conclude that the solutions on the branch `` 1 '' are unstable ( they are sphaleron - like ) and it is very likely that the monopole looses its non - trivial topology and its status of being a local minimum when it is coupled to the other fields appearing in the 4-dimensional effective model .
there are @xmath99 negative -unstable- modes while the solution of branch `` 2 '' have @xmath100 unstable modes .
a direct analysis of the stability should definitely confirm this conclusion and provide the value of @xmath99 .
the dimensional reduction of the einstein - yang - mills theory from @xmath2 dimensions down to the standard @xmath0 dimensions is an interesting problem .
the main source of uncertainty is the way the fields depend on the extra dimensions .
one can adopt , for the simplest case , the point of view that , in an appropriate gauge and with appropriate variables , the gauge and metric fields are independent on the extra coordinates .
a 4-dimensional effective action which encodes all the effects of the extra dimension into a more or less `` conventional '' field theory can then be constructed . in the case of @xmath101
, it was shown that the effective action is a georgi - glashow model appropriately coupled to one dilaton field @xcite .
here we have studied in details the case @xmath102 .
one of the main differences in comparison to the @xmath101 case is that the einstein - yang - mills equations are consistent only if the metric has off - diagonal terms in the subspace of the extra dimensions .
the corresponding effective action is a standard einstein - yang - mills action supplemented by two higgs triplets and three scalar fields , originating from the parametrization of the metric in the codimensions .
the function corresponding to the off - diagonal term of the metric plays a very important role in the effective action , largely determining an interaction between the two higgs fields .
all these fields acquire kinetic parts as well as specific interactions .
several possibilities of extension of the result can be investigated , namely ( 1 ) the construction of non - abelian black hole , ( 2 ) the stability analysis of the different branches of regular solutions , ( 3 ) the addition of higher order terms in the yang - mills action ( `` born - infeld '' terms ) and ( 4 ) the construction of axially symmetric solutions .
99 t. kaluza , sitzungsber .
berlin ( 1921 ) . o. klein , z. phys .
* 37 * ( 1926 ) , 895 .
g. gibbons and k. maeda , nucl .
b298 * ( 1988 ) , 741 ; d. garfinkle , g. horowitz and a. strominger , phys . rev . * d43 * ( 1991 ) , 371 .
see e.g. j. polchinski , _ string theory _
, cambridge university press ( 1998 ) .
k. akama , `` pregeometry '' , in lecture notes in physics , 176 , gauge theory and gravitation , proceedings , nara , 1982 , edited by k. kikkawa , n. nakanishi and h. nariai , 267 - 271 ( springer - verlag , 1983 ) [ hep - th/0001113 ] ; v. a. rubakov and m. e. shaposhnikov , phys
* 125 b * ( 1983 ) , 136 ; * 125 b * ( 1983 ) , 139 ; g. davli and m. shifman , phys .
* b 396 * ( 1997 ) , 64 ; * 407 * ( 1997 ) , 452 ; i. antoniadis , phys . lett .
* b 246 * ( 1990 ) , 377 ; n. arkani - hamed , s. dimopoulos and g. dvali , phys . lett .
* b 429 * ( 1998 ) , 263 ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. dvali , phys . lett .
* b 436 * ( 1998 ) , 257 ; l. randall and r. sundrum , phys .
* 83 * ( 1999 ) , 3370 ; * 83 * ( 1999 ) , 4690 .
m. s. volkov , phys .
b524 * ( 2002 ) , 369 .
y. brihaye , f. clement and b. hartmann , _ spherically symmetric yang - mills solutions in a ( @xmath2)-dimensional space - time _ , hep - th/0403041
t hooft , nucl .
phys . * b79 * ( 1974 ) , 276 ; + a. m. polyakov , jetp lett .
* 20 * ( 1974 ) , 194 . | we study the einstein - yang - mills equations in a 6-dimensional space - time .
we make a self - consistent static , spherically symmetric ansatz for the gauge fields and the metric .
the metric of the manifold associated with the two extra dimensions contains off - diagonal terms .
the classical equations are solved numerically and several branches of solutions are constructed .
we also present an effective 4-dimensional action from which the equations can equally well be derived .
this action is a standard einstein - yang - mills - higgs theory extended by three scalar fields .
two of the scalar fields are interpreted as dilatons , while the one associated with the off - diagonal term of the metric induces very specific interactions . |
strong electron correlations in multi - band , orbitally degenerate systems represent an important current theoretical challenge . a lively experimental playground for that is provided by electron- and hole - doped fullerene systems , which exhibit a variety of behavior , including unconventional metals like cubic csc@xmath6@xcite , superconductors of the a@xmath5c@xmath6 family ( a= k , rb , cs)@xcite , superconducting fet devices@xcite , and insulators , presumably of the mott - jahn - teller type@xcite . among them we can place a@xmath7c@xmath6 @xcite , na@xmath8c@xmath6@xcite and the class of compounds nh@xmath5k@xmath5c@xmath6 , nh@xmath5k@xmath8rbc@xmath6 , nh@xmath5krb@xmath8c@xmath6 , as well as nh@xmath5rb@xmath5c@xmath6@xcite . in these ammoniated compounds ,
insertion of the electronically inert nh@xmath5 molecules acts to expand the c@xmath6 lattice , turning the cubic , metallic and superconducting state of k@xmath5c@xmath6 into an orthorhombic narrow - gap antiferromagnetic insulator @xcite . in this transition ,
the increase in volume per c@xmath6 molecule relative to k@xmath5c@xmath6 , with its probable slight decrease of electron effective bandwidth @xmath9 , is believed to play an important role , as confirmed by the pressure - induced reversion to a fully metallic and superconducting while still orthorhombic
phase @xcite . nonetheless , a possible role of the crystal - structure anisotropy in this mott transition a transition generally believed to occur more readily for reduced orbital degeneracy@xcite , can not be excluded .
it has even been proposed that precisely the splitting of degeneracy@xcite induced by the orthorhombic distortion could be the crucial ingredient driving the transition from the metal and superconductor to the insulator @xcite .
earlier work showed the mott transition in @xmath10-orbitally degenerate lattice models to take place at larger values of @xmath11 for larger degeneracy , roughly proportionally to @xmath12 @xcite .
a strong enough `` crystal - field '' splitting of the threefold degenerate @xmath13 molecular orbital of c@xmath6 caused by orthorhombic anisotropy could remove one or several bands away from the fermi level , effectively reducing the orbital degeneracy @xmath10 , thus shifting the mott transition to a smaller critical value @xmath14 . exploring this concept ,
the basic question is how strong this splitting must be to promote such an effective reduction of degeneracy . in a non - interacting system
, that reduction would clearly require a splitting magnitude similar to or larger than the full electron bandwidth @xmath9 . in a strongly - interacting system , it is important to understand whether influencing the metal - insulator transition will again require anisotropic splittings as large as the bandwidth , or else if some smaller energy scale will emerge in its place . in this paper
we exploit dynamical mean field theory ( dmft)@xcite to answer this question , studying the effects of a band splitting on the mott transition of an orbitally degenerate , strongly - correlated metal . for that purpose
we adopt the simplest idealization , namely a @xmath15 orbitally degenerate hubbard model with bandwidth @xmath9 and on - site repulsion @xmath16 , and mimic the noncubic anisotropy by a diagonal splitting @xmath0 between the two bands . in this model , we study the @xmath17 phase diagram in the @xmath18 plane , and examine in particular how the mott transition is influenced by the splitting . as the aim of the calculation is understanding a simple , and possibly fairly general physical mechanism , we do not attempt a realistic calculation of the kind that was accomplished recently for ni and fe @xcite .
the main result found is that the effect is remarkably strong at large @xmath16 , where the anisotropy - induced transition between the correlated multi - band metal and the mott insulator takes place very readily , in fact for @xmath19 .
we interpret that as follows . the metallic band , originally of width @xmath9 , is renormalized in the strongly correlated metal into a narrower `` kondo - like '' quasiparticle peak of width @xmath3 , with @xmath20 as @xmath21 .
close enough to the isotropic and orbitally degenerate mott transition at @xmath22 , the correlated two - band metal can be driven effectively one band by a splitting as small as the kondo energy scale @xmath23 .
the one - band metal , in turn , has a lower @xmath24 , so that the resulting state is an insulator . with this model result in hand ,
we move on to consider nh@xmath5k@xmath5c@xmath6 , where we carry out a dft - lda ( density functional theory in the local density approximation ) electronic - structure calculation to extract an order of magnitude for the effective orthorhombic splitting parameter @xmath25 , found to be of order @xmath26 . despite the obvious differences between the two - band hubbard model and the true three band system
, we suggest that the orthorhombic splitting will similarly drive the initially cubic fulleride with a small quasiparticle residue @xmath27 from metal to insulator . in turn , in the light of our simple - model phase diagram , the return to metallicity of the ammoniated orthorhombic fulleride under pressure leads to an interesting question regarding the effective number of bands in that metal .
it is finally concluded that anisotropy is likely a relevant element in promoting the mott state in this material .
this paper is organized as follows : sec .
[ model : sec ] introduces the model and deals with its simple limiting cases .
the calculation and results within the dmft are described is sec .
[ dmft : sec ] .
the @xmath13 band structure of nh@xmath5k@xmath5c@xmath6 and its splitting due to anisotropy is described in sec .
[ bands : sec ] . the discussion and conclusions are finally presented in sec .
[ discussion : sec ] .
we work with the simplest two - band hubbard model , assuming purely diagonal hoppings between orbitals on different sites , the anisotropic symmetry lowering all embodied in a diagonal on - site splitting term .
we write the hamiltonian as : @xmath28 where @xmath29 creates a spin-@xmath30 electron in orbital @xmath31 at site @xmath32 , @xmath33 , @xmath34 is the total number of electrons at site @xmath32 , and @xmath35 denotes nearest - neighbor sites .
the first term is a standard tight - binding hopping , simplified to ignore non - diagonal hopping , and also merohedral disorder , both of them present in the real systems .
the second term introduces the anisotropic splitting @xmath0 as a shift of the on - site orbital energy @xmath36 .
the last term is the hubbard on - site interaction , which we take in its simplest form , omitting for example any additional intra - site hund s rule interaction terms .
we also ignore the on - site electron - vibron couplings and the ensuing jahn - teller effects , even if quite important for other aspects@xcite . in the @xmath37 limit ,
the splitting @xmath38 simply operates a rigid shift of band 2 upwards with respect to band 1 , promoting electron transfer from the upper to the lower band . above a critical value @xmath39 , and for a total electron filling @xmath40 , the upper band is emptied up .
for example , with two symmetric bands of width @xmath9 and @xmath41 electrons per site , the upper band is emptied above @xmath42 . at zero temperature ( where we shall restrict ourselves in this paper ) and zero interaction , @xmath37 , the transition between the `` two - band metal '' and
the `` one - band metal '' is continuous . because the topology of the fermi surface changes , this transition is accompanied by a weak singularity of the total energy first described by lifshitz@xcite .
when the electron - electron interaction @xmath16 is turned on , one expects the emptying of the upper band to take place at smaller values of @xmath43 , owing to the effective band narrowing .
perturbatively in @xmath16 one can show that @xmath44 where the value of the coefficient @xmath45 depends on details of the bands .
in addition , electron - electron interactions might modify the nature of the metal - metal transition singularity relative to the non - interacting case@xcite , a point which we will not further address here .
we restrict to a filling of one electron per site , @xmath46 , discarding the trivial case @xmath47 where the large-@xmath0 state is a band insulator , as well as noninteger @xmath48 , where insulators are not possible .
the @xmath49 and @xmath50 limits reduce then , respectively , to the ( quarter - filled ) two - band and the ( half - filled ) single - band hubbard models , both possessing a metal - insulator transition as a function of @xmath16 @xcite .
in particular , we are not interested in the weak - coupling antiferromagnetic instability , of the ensuing large-@xmath0 half - filled band , associated to nesting in specific hopping structures : we assume a genuine mott transition for the half - filled single - band model to occur at a finite @xmath51 .
the limit of strong interaction @xmath52 is insulating for any value of @xmath0 .
this limit can be studied by mapping the model ( [ hamiltonian : eq ] ) onto a spin and orbital exchange hamiltonian which reads@xcite : @xmath53 where the pseudospin-1/2 operators @xmath54 represent the orbital degrees of freedom , and @xmath55 . for @xmath49 this model has been studied both in one@xcite and two dimensions @xcite , with suggestions that interesting spin - liquid physics could be realized . for our purposes , it suffices to note that the model has no ferro - orbital instability , and has therefore a finite @xmath56 orbital susceptibility . as a consequence it takes a nonzero value of @xmath0 to fully orbitally polarize the ground state . due to
the absence of cross - band terms in the kinetic energy , complete orbital polarization occurs at a finite @xmath57 . for @xmath58 that ground state
is represented by a one band mott insulator plus a totally empty split - off band .
notice , finally , that for the specific case we are interested ( @xmath46 electron in a two - fold degenerate band ) no orbital ordering is present at weak coupling ; the possibility of antiferro - orbital ordering within the mott insulating phase , on the other hand , depends crucially on the details of the hoppings and lattice coordination , and it is , in any case , beyond the scope of any single - site dmft study .
we have now all the elements to illustrate the general layout , sketched in fig .
[ phased_th : fig ] , of the zero - temperature phase diagram of model in eq .
( [ hamiltonian : eq ] ) as a function of @xmath18 for @xmath46 electron per site .
the metal - insulator transition for @xmath48=1 , @xmath15 and @xmath49 , previously studied in dmft by quantum monte carlo @xcite , was found at @xmath59 at the relatively low temperature @xmath60 .
the same transition with @xmath46 in the one - band model ( @xmath61 , or @xmath15 at @xmath62 ) has been similarly calculated at @xmath63 also at @xmath60@xcite .
finite temperature of course must affect somewhat these numerical values , but a common feature with our zero - temperature dmft calculation described below , is that @xmath64 . the ab line in fig .
[ phased_th : fig ] , representing @xmath24 as a function of @xmath0 and separating metals from mott insulators , indicates this trend .
the dm line separates the fully orbitally polarized mott insulator ( at the right ) from the two - band insulator , roughly as @xmath65 for small @xmath0 .
similarly , the cm line separates the fully orbitally polarized metal from the two - band metal : it starts from point c with a linear slope @xmath66 , according to eq .
( [ smallu ] ) . in the region of full orbital polarization
the value of @xmath0 is irrelevant , and this is the reason why the mott transition line mb is horizontal . as @xmath0 increases , for @xmath67 the upper - band emptying transition takes the two - band metal across the cm line over to a one - band metal , while for @xmath68 it leads across the am line to a mott insulating state .
the am line starts off linearly at small @xmath0 .
basically , this occurs since in close proximity of the mott transition the relevant energy scale , to be matched by the orbital splitting @xmath0 , is no longer the bare bandwidth , but rather the width of the kondo - like quasiparticle peak of the correlated metal .
that in turn is of the order of @xmath3 , which is smaller than @xmath9 and proportional to the quasiparticle residue @xmath27 , linearly vanishing with @xmath69 at the mott transition@xcite . as the dmft calculations will show , the effective emptying transition occurs when @xmath0 increases to reach @xmath70 .
the purpose of the dmft calculations is to investigate this phase diagram in greater quantitative detail , in particular in the region of intermediate @xmath71 and @xmath72 .
we have implemented dmft in the exact diagonalization flavor @xcite for the degenerate hubbard model on the bethe lattice , restricted to paramagnetic states @xcite .
the advantage of the exact diagonalization method over the monte carlo approach in the solution of the impurity problem rests primarily in its being less computationally intensive
useful when studying the phase diagram of a model as a function of more than one parameter and in its ability to address zero - temperature properties @xcite .
the results of the dmft calculations are summarized in fig .
[ phase_diagram : fig ] .
the values we obtain for @xmath73 and @xmath74 , associated to the disappearance of the metallic state for one and two bands respectively , have been calculated with the same number of discretized bath levels per band @xmath75 .
the associated error bar is estimated by repeating the calculation with @xmath76 .
these values are in substantial agreement with corresponding values obtained by other methods@xcite .
the other points in the phase diagram are obtained by following the stability of the two - band metal for a given value of @xmath16 and increasing @xmath0 , marking the emptying transition to the one - band metal or to the insulator .
a deficiency of this single - site dmft calculation which is restricted as usual to paramagnetic states only is the absence of a two - band insulating state for @xmath38 .
in fact , the suppression of the antiferro orbital fluctuations embodied in the exchange model ( [ hexch ] ) produces a fictitious infinite uniform orbital susceptibility which leads to full orbital polarization as soon as @xmath0 is turned on @xcite . despite this limitation ,
the results of the present calculations are suggestive , revealing the announced sharp reduction of the metal - insulator @xmath77 for small but finite @xmath0 . indeed
, our calculations give a @xmath78 which is roughly proportional to the quasiparticle residue @xmath79 of the undistorted ( @xmath49 ) correlated metal at @xmath80 , @xmath81 with a proportionality constant @xmath82 .
since @xmath79 vanishes as @xmath83 , most likely linearly in @xmath69 @xcite , even a very small @xmath25 can be enough to cause a metal - insulator transition in the strongly correlated metal .
for example , following the bold arrow at @xmath84 in fig .
[ phase_diagram : fig ] , a @xmath0 value as small as @xmath85 is sufficient to cross the transition line from the metal to insulator . further illustrating that , in fig . [
spec_dens : fig ] we show the behavior of the spectral density @xmath86 @xmath87 being the one - particle green s function of band @xmath31 , on both sides of the metal - insulator transition .
the asymmetry of the upper - band spectral density @xmath88 ( solid lines ) is very pronounced , as this band is nearly ( @xmath89 ) or completely ( @xmath90 ) empty .
as soon as the kondo - like peaks of the two bands differ enough in energy to induce the emptying of band 2 , the lower - band spectral density @xmath91 takes the symmetric shape , characteristic of the half - filled one - band hubbard model . here
the kondo peak disappears completely , as this value of @xmath92 puts the hubbard model of band 1 well inside the insulating regime .
the above model calculations show that a small splitting @xmath93 of the orbitally degenerate band can drive the metal - insulator transition .
we now wish to explore the implications that this result if assumed to be more general than the simple model where it was derived can have on the metal - insulator transition which takes place between isoelectronic k@xmath5c@xmath6 and nh@xmath5k@xmath5c@xmath6 ( the former cubic and the latter orthorhombic ) , and on the insulator - metal transition of nh@xmath5k@xmath5c@xmath6 itself under pressure . to obtain a quantitative estimate of the actual strength of anisotropy in the ammoniated fulleride , we carried out comparative dft - lda electronic structure calculations for k@xmath5c@xmath6 and a simplified but meaningful model of nh@xmath5k@xmath5c@xmath6 using the pwscf package based on plane waves and ultrasoft pseudopotentials@xcite .
we took a lattice constant @xmath94 for fcc k@xmath5c@xmath6 , and @xmath95 for nh@xmath5k@xmath5c@xmath6 , the latter with a centered tetragonal unit cell with @xmath96 , neglecting the exceedingly small difference between @xmath97 and @xmath98 . in both calculations a single - c@xmath6 unit cell was used , as pictured in fig .
[ geometry_nh3:fig ] . in these model structures and calculations , we ignored merohedral disorder , as well as the rich antiferro - rotational structure recently discovered in actual nh@xmath5k@xmath5c@xmath6@xcite .
we used 27 ry and 160 ry plane wave energy cut - offs for the wavefunctions and the charge density , respectively .
the fermi surface was smeared with parameter of 0.3 ev and a @xmath99 monkhorst and pack mesh @xcite .
the density of states ( dos ) was calculated as a sum of gaussians of width @xmath100 ev sampling @xmath101-space with a uniform @xmath102 mesh .
[ bands_nh3:fig ] shows the @xmath13 bands of nh@xmath5k@xmath5c@xmath6 ( a ) , compared to those of k@xmath5c@xmath6 ( c ) ; the corresponding density of states are compared in panels ( b ) and ( d ) .
we find that the presence of the inserted nh@xmath5 molecules modifies only weakly the essentially pure c@xmath6 @xmath13 conduction band , as expected .
the difference in the two band structures are almost entirely due to the different lattice structures of the molecular centers .
small changes appear in the details of the bands throughout the brillouin zone , but both compounds are predicted in lda to be three - band metals , their overall @xmath13 density of states not dissimilar in the ammonia - intercalated and pristine compound . in particular , as apparent from fig .
[ bands_nh3:fig](b ) , the bandwidths of nh@xmath5k@xmath5c@xmath6 and k@xmath5c@xmath6 are close , @xmath103 ev . the main difference in the two band structures is a splitting of the threefold degenerate @xmath13 band of k@xmath5c@xmath6 at the @xmath104 point of nh@xmath5k@xmath5c@xmath6 .
this splitting is precisely a measure of the strength of the non - cubic crystalline environment seen by the @xmath13 orbital on each fullerene molecule in the orthorhombic structure of nh@xmath5k@xmath5c@xmath6 .
its magnitude is roughly a quarter ( 0.2 - 0.3 ) of the total bandwidth ; this represents the main result of the dft calculation .
if this material were an uncorrelated metal , this splitting would have no major consequences , just a change of shape of the fermi surface .
the consequences can be much more important due to strong correlations , as discussed below .
the @xmath104-point @xmath13 band splitting see fig . [
bands_nh3:fig](a ) can be taken as a crude estimate of the value of @xmath105 ev in nh@xmath5k@xmath5c@xmath6 , corresponding to a dimensionless ratio @xmath106 .
this value is of course too small to determine a complete emptying of one of the bare , uncorrelated bands , and indeed all three bands of nh@xmath5k@xmath5c@xmath6 still cross the fermi energy @xmath107 .
however , correlations play an important role in fullerides , where the value of @xmath16 is believed to lie between 1 and 1.6 ev @xcite , thus substantially larger than the bandwidth @xmath9 . as the non - ammoniated fullerides a@xmath5c@xmath6 are still generally in the metallic and superconducting phase , it is not unreasonable to assume that their value of @xmath11 @xcite is in all likelihood only marginally smaller than the @xmath108 for the mott transition in the half - filled @xmath109-band cubic system .
indirect evidence that ordinary , uncorrelated electronic bands do not exist , and that the electron spectral function at @xmath107 can more reasonably be assimilated to a narrow , dispersionless kondo - like resonance was provided for k@xmath5c@xmath6 by recent photoemission data@xcite .
the analysis of the metal - insulator transition in our simplified two - band model suggests a mechanism for an anisotropy - induced mott transition in a correlated multi - band fermionic system .
clearly , many ingredients which we have left out of our calculation are expected to be quantitatively important for a realistic description of the @xmath13 doped fullerides .
however , we suggest that the mechanism which we have described in secs . [ model : sec ] and [ dmft : sec ] , whereby a small orthorhombic splitting of the bands is capable , alone , of turning a strongly correlated metal into a mott insulator , could be a qualitatively important ingredient in the description of these systems .
specifically , we suggest that , in the appropriate @xmath18 phase diagram ( sketched in fig . [ phased_nh3:fig ] ) , the ammoniation of k@xmath5c@xmath6 into orthorhombic nh@xmath5k@xmath5c@xmath6 could correspond to a displacement@xcite similar to that indicated by an arrow in fig .
[ phase_diagram : fig ]
. inspection of fig .
[ phase_diagram : fig ] shows that , if @xmath16 is close enough to @xmath24 , even a value of @xmath0 substantially smaller than that we have estimated for nh@xmath5k@xmath5c@xmath6 could suffice to drive that metal insulator transition .
very recent conductivity measurements @xcite on the class of compounds nh@xmath5k@xmath110rb@xmath111c@xmath6 suggest indeed the orthorhombic distortion as the main ingredient driving the mott transition in these systems .
we should stress that our picture does not exclude a role , in the metal - insulator transition , for the accompanying ammoniation - driven expansion of the lattice . that role is experimentally proven by the pressure - driven insulator - metal transition , where nh@xmath5k@xmath5c@xmath6 is transformed into a metal and a superconductor , despite the permanence of the orthorhombic structure@xcite . however a new , interesting , question raised in fig .
[ phased_nh3:fig ] is what kind of metal does one recover in nh@xmath5k@xmath5c@xmath6 under pressure , whether a three - band metal like the cubic k@xmath5c@xmath6 , or a new effective two - band metal .
the former might in effect require a smaller deformation .
it would be very interesting in this regard if the fermi - surface topology of nh@xmath5k@xmath5c@xmath6 could be studied under pressure .
another interesting even if perhaps not practically straightforward test of this overall picture could be obtained by applying uniaxial stress to cubic superconducting fullerides of the a@xmath5c@xmath6 family .
contrary to standard tendency of hydrostatic pressure toward metallization and lower superconducting @xmath112 , the orbital splitting associated with the appropriate uniaxial strain could drive some of these compounds perhaps first toward higher @xmath112 , and eventually mott insulating .
a similar suggestion has been recently put forward , in the context of field - induced superconductivity @xcite , by koch @xcite , although on rather different grounds .
also , recent calculations @xcite of the size and the effects of the charge - inducing electric field on the 2-dimensional c@xmath6 layer , in the field - induced geometry , suggest that this field induces splittings of the molecular levels which might in turn affect the electrons and holes bands in a qualitatitevely similar way to the one we propose here for the orthorhombic field in 3-dimensional fullerides . in summary
, we have investigated the role of a symmetry - breaking orbital splitting on the transition to a mott insulating state in an initially orbitally degenerate strongly correlated multi - band metal .
the calculations , carried out in a highly idealized @xmath15 model , have unveiled the kondo quasiparticle energy scale as that which the non - cubic splitting magnitude must reach in order to induce the mott transition even at constant bandwidth and effective coulomb repulsion .
the likely relevance of this main qualitative finding to the fullerides k@xmath5c@xmath6 and nh@xmath5k@xmath5c@xmath6 has been highlighted .
we are indebted to o. gunnarsson and m. fabrizio , for useful discussions .
r. assaraf collaborated with this project in its earliest stage .
this work was partly supported by the european union , contract erbfmrxct970155 ( tmr fulprop ) , and by miur cofin01 , murst cofin99 .
working at zero temperature is a quite important feature in our specific case , where the competition between the splitting @xmath0 and the small quasi - particle residue @xmath27 might be blurred , within a standard monte carlo approach , by finite temperature effects .
the exact diagonalization method suffers , however , from a severe constraint on the number of discretized conduction band states which can be included . as a consequence , for realistic calculations with three or more degenerate orbitals
a monte carlo approach is more appropriate @xcite . | within dynamical mean - field theory , we study the metal - insulator transition of a twofold orbitally degenerate hubbard model as a function of a splitting @xmath0 of the degeneracy .
the phase diagram in the @xmath1 plane exhibits two - band and one - band metals , as well as the mott insulator .
the correlated two - band metal is easily driven to the insulator state by a strikingly weak splitting @xmath2 of the order of the kondo - peak width @xmath3 , where @xmath4 is the metal quasiparticle weight .
the possible relevance of this result to the insulator - metal transition in the orthorhombic expanded fulleride nh@xmath5k@xmath5c@xmath6 is discussed . |
the current astrophysical observations of the type ia supernovae ( snia ) @xcite , the cosmic microwave background ( cmb ) @xcite and the large scale structure ( lss ) @xcite have revealed that the expansion of our universe is accelerated @xcite .
this indicates that there exists some unknown energy , called dark energy , to realize the accelerated expansion .
the simplest interpretation of dark energy is the cosmological constant .
however , this model requires an incredible fine - tuning , since the observed cosmological constant is extremely small compared to the fundamental planck scale @xmath4 .
also , this model suffers from the cosmic coincidence problem : why the cosmological constant and matter have comparable energy density today even though their time evolution is so different . among various attempts to solve these problems ,
we focus on the holographic dark energy ( hde ) models @xcite motivated by the holographic principle of quantum gravity @xcite . requiring that the total vacuum energy of a system with size @xmath5 would not exceed the mass of the black hole of the same size
, the dark energy density is proposed as @xmath6 where @xmath7 is a dimensionless parameter , @xmath8 is the reduced planck mass . as for the size @xmath5 , which is regarded as an ir cutoff , various possibilities are discussed in literatures , such as the hubble parameter @xmath9 @xcite , the future event horizon @xmath10 @xcite , the age of our universe @xmath11 @xcite , and the ricci scalar curvature @xmath12 @xcite . in our previous work
@xcite , we studied the ricci dark energy ( rde ) model with @xmath12 by introducing an interaction between dark energy and matter .
it was shown that a nonvanishing interaction rate @xmath13 is favored by the observations @xcite . in this paper , we extend our previous analysis , and consider the interacting rde ( irde ) model in non - flat universe .
this paper is organized as follows . in sec .
[ irde ] , we describe the generalized irde model in non - flat universe , and obtain analytic expressions for cosmic time evolution . in sec .
[ obs ] , we discuss the observational constraints on this model .
we summarize our results in sec .
we study the interaction ricci dark energy ( irde ) model in non - flat universe .
the friedmann - robertson - walker metric non - flat univrerse is given by @xmath14 where @xmath15 , 0 , @xmath16 for closed , flat , and open geometries .
the friedmann equation in non - flat univrerse takes the form @xmath17 where @xmath18 and @xmath19 represent energy density of dark energy , matter , radiation and curvature , respectively , and @xmath20 is the hubble parameter .
we generalize the energy density of the ricci dark energy as @xmath21 where @xmath22 , @xmath23 and @xmath24 are dimensionless parameters and @xmath25 . in the case of @xmath26 , this model
is reduced to the ordinary rde model @xcite .
moreover , we introduce a phenomenological interaction between dark energy and matter .
the energy densities @xmath27 and @xmath28 obey the following equations @xcite @xmath29 we adopt the interaction rate given by @xmath30 where @xmath31 is a dimensionless parameter @xcite . to solve @xmath32 , combining with eqs .
( [ rde ] ) and ( [ clm ] ) , the friedmann equation ( [ friedmann ] ) is transformed as @xmath33 the solution to eq .
( [ eoh ] ) is given by @xmath34 where @xmath35 @xmath36 , @xmath37 and @xmath38 .
when @xmath39 can be imaginary for sufficiently large @xmath22 , @xmath23 and @xmath31 , @xmath40 has oscillatory behavior @xcite .
the constants @xmath41 , @xmath42 and @xmath43 are obtained as @xmath44 @xmath45 @xmath46 in the case of @xmath47 , @xmath48 and @xmath49 , eq . ( [ solution ] ) reduces to the result obtained in our previous work @xcite .
substituting eq .
( [ solution ] ) to eq .
( [ rde ] ) , the ricci dark energy density is obtained as @xmath50 where @xmath51 , @xmath52 , likewise , the matter density is obtained as @xmath53 to derive the equation of state parameter @xmath54 of the ricci dark energy , substituting eq .
( [ rdes ] ) into the following expression : @xmath55
in this section , we discuss cosmological constraints on the irde model in the non - flat universe ( @xmath56 ) obtained from snia , cmb , bao and the hubble parameter observations .
the luminosity distance in the non - flat universe can be written as @xmath57 the snia observations measure the distance modulus @xmath58 of a supernova and its redshift @xmath59 .
the distance modulus is given by @xmath60 we use the union data set of 580 snia @xcite to obtain limits on the relevant parameters @xmath61 and @xmath62 by minimizing @xmath63 @xcite .
the cmb shift parameter @xmath64 is given by @xmath65 where @xmath66 is the redshift at recombination , and @xmath67 is the matter fraction at present .
we use the value @xmath68 obtained from the wmap9 data @xcite .
the cmb constraints are obtained by minimizing @xmath69 @xcite .
the shift parameter gives a complementary bound to the snia data ( @xmath70 ) , since this parameter involves the large redshift behavior ( @xmath71 ) .
signatures of the baryon acoustic oscillation ( bao ) are provided by the observations of large - scale galaxy clustering .
the bao parameter @xmath72 is defined by @xmath73^{2/3 } , \end{aligned}\ ] ] where @xmath74 .
we use the measurement of the bao peak in the distribution of luminous red galaxies ( lrgs ) observed in sdss @xcite : @xmath75 the bao constraints are obtained by minimizing @xmath76 @xcite .
the hubble parameter constraints are given by minimizing @xmath77 where @xmath78 is the @xmath79 uncertainty of the observational @xmath80 data @xcite .
in fig.[fig : w_a_gamma ] , we plot the equation of state parameter @xmath54 for @xmath81 ( blue ) , 0 ( red ) and 0.1 ( green ) as a function of the scale factor @xmath82 .
the dotted and solid lines are the results for the case without interaction ( @xmath31=0 ) and with interaction ( @xmath83 ) , respectively .
the dark energy parameter is fixed as @xmath84 . though the value @xmath85 is too large ( see fig .
[ fig : contours ] ) , we use these values to visualize the effect of the curvature in the figure . the condition for accelerated expansion at present @xmath86 @xmath87 @xmath88 is satisfied for both cases . in fig.[fig :
omega_a_gamma ] , we plot the evolution of the energy density fractions @xmath89 for radiation ( green ) , matter ( red ) and dark energy ( blue ) for @xmath84 .
the dotted and solid lines are the results for the case without interaction ( @xmath31=0 ) and with interaction ( @xmath83 ) , respectively .
the panels ( a ) , ( b ) and ( c ) corresponds to @xmath49 , @xmath90 and 0.1 , respectively .
one can see that the effect of the curvature can be important only for @xmath82 @xmath91 1 .
the fractions @xmath92 and @xmath93 for @xmath81 in the panel ( b ) ( @xmath94 in the panel ( c ) ) are slightly increased ( decreased ) around 0.1 @xmath1 @xmath82 @xmath1 1 compared to the flat case ( a ) . around radiation - matter equality , the radiation component @xmath95 is increased by several percents compared to the corresponding result without interaction ( dotted line ) , while the @xmath93 is decreased due to the interaction . ) -plane in the case without interactions ( @xmath96 ) .
the @xmath79 , @xmath97 and @xmath98 contours are drawn with solid , dashed and dotted lines , respectively .
the joined constraints using @xmath99 are shown as shaded contours .
[ fig : contours ] , title="fig:",scaledwidth=95.0% ] + ( a ) ) -plane in the case without interactions ( @xmath96 ) .
the @xmath79 , @xmath97 and @xmath98 contours are drawn with solid , dashed and dotted lines , respectively .
the joined constraints using @xmath99 are shown as shaded contours .
[ fig : contours ] , title="fig:",scaledwidth=95.0% ] + ( b ) versus @xmath82 for rde(@xmath100 ) the lines for k=-0.1 ( blue ) , 0 ( red ) and 0.1 ( green ) in the case without interactions ( dotted line ) and with interaction ( solid line ) . ,
width=364 ] versus @xmath82 for rde(@xmath101 ) .
the lines for radiation ( green ) , matter ( red ) , dark energy ( blue ) and curvature ( black ) in the case without interactions ( dotted line ) and with interaction ( solid line ) .
these figures describe in the case of @xmath49 ( fig .
a ) , -0.1 ( fig .
b ) and 0.1 ( fig .
[ fig : omega_a_gamma ] , width=264 ] ( a ) versus @xmath82 for rde(@xmath101 ) . the lines for radiation ( green ) , matter ( red ) , dark energy ( blue ) and curvature ( black ) in the case without interactions ( dotted line ) and with interaction ( solid line ) .
these figures describe in the case of @xmath49 ( fig .
a ) , -0.1 ( fig .
b ) and 0.1 ( fig .
[ fig : omega_a_gamma ] , width=268 ] ( b ) versus @xmath82 for rde(@xmath101 ) . the lines for radiation ( green ) , matter ( red ) , dark energy ( blue ) and curvature ( black ) in the case without interactions ( dotted line ) and with interaction ( solid line ) .
these figures describe in the case of @xmath49 ( fig .
a ) , -0.1 ( fig .
b ) and 0.1 ( fig .
c ) . [ fig : omega_a_gamma ] , width=264 ] ( c )
we have considered the irde model in the non - flat universe .
we have derived the analytic solutions for the hubble parameter ( [ hubbleparameter ] ) , the dark energy density ( [ rdes ] ) and matter energy density ( [ rms ] ) .
we have also studied astrophysical constraints on this model using the recent observations including snia , bao , cmb anisotropy , and the hubble parameter .
we have shown that the allowed parameter range for the fractional energy density of the curvature is @xmath0 @xmath1 @xmath2 @xmath1 @xmath3 for @xmath31 @xmath102 0.15 .
the best fit values with @xmath79 error are @xmath103 and @xmath104 with @xmath105 .
we have shown that the irde model with a small curvature is allowed by observational constraints . without the interaction , the flat universe
is observationally disfavored in this model .
we would like to thank t. nihei for his valuable discussion , helpful advice and reading the manuscript .
99 a. g. riess et al .
j. 116 ( 1998 ) 1009 ; s. perlmutter et al .
, astrophys .
j. 517 ( 1999 ) 565 . c. l. bennet et al .
, astrophys .
j. suppl . 148
( 2003 ) 1 ; d. n. spergel et al .
, astrophys .
j. suppl . 148
( 2003 ) 175 ; d. n. spergel et al .
, astrophys . j. suppl . 170
( 2007 ) 377 ; l. page et al .
, astrophys .
j. suppl . 170
( 2007 ) 335 ; g. hinshaw et al .
, astrophys .
j. suppl . 170
( 2007 ) 263 ; e. komatsu et al .
, astrophys .
j. suppl . 180
( 2009 ) 330 [ arxiv:0803.0547 ] ; e. komatsu et al .
[ wmap collaboration ] , astrophys .
192 , 18 ( 2011 ) ; g. hinshaw , d. larson , e. komatsu , d. n. spergel , c. l. bennett , j. dunkley , m. r. nolta and m. halpern et al . , arxiv:1212.5226 [ astro-ph.co ]
d. j. eisenstein et al . , astrophys .
j. 633 ( 2005 ) 560 [ arxiv : astro - ph/0501171 ] .
m. li , phys .
b 603 ( 2004 ) 1 .
g. t hooft , arxiv : gr - qc/9310026 ; l. susskind , j. math .
36 ( 1995 ) 6377 [ arxiv : hep - th/0403052 ] ; j. m. maldacena , adv . theor . math .
phys . 2 ( 1998 ) 231 ; w. fischler , l. susskind , arxiv : hep - th/9806039 .
a. einstein , sitzungsber .
berlin ( math .
phys . ) 1917 ( 1917 ) 142 .
see e.g. , s. weinberg , astro - ph/0005265 .
a. cohen , d. kaplan , a. nelson , phys .
82 ( 1999 ) 4971 [ arxiv : hep - th/9803132 ] .
s. d. h. hsu , phys .
b 594 ( 2004 ) 13 [ arxiv : hep - th/0403052 ] .
cai , phys .
b 657 , 228 ( 2007 ) ; c. gao , f. wu , x. chen , y. g. shen , phys . rev .
d 79 ( 2009 ) 043511 [ arxiv:0712.1394 ] . l. xu , j. lu , w. li , eur . phys .
j. c 64 ( 2009 ) 89 [ arxiv : astro - ph/0906.0210 ] ; c. j. feng , x. z. li , phys . lett .
b 680 ( 2009 ) 355 ; c. j. feng , x. z. li , phys .
b 679 ( 2009 ) 151 [ arxiv : hep - th/0904.2976 ] ; s. b. chen , j. l. jing , arxiv:0904.2950 ; c. j. feng , x. zhang , phys . lett .
b 680 ( 2009 ) 399 [ arxiv : gr - qc/0904.0045 ] .
b. wang , c. y. lin , e. abdalla , phys .
b 637 ( 2006 ) 357 [ hep - th/0509107 ] ; m. s. berger , h. shojaei , phys . rev .
d 74 ( 2006 ) 043530 [ astro - ph/0606408 ] ; m. r. setare , phys . lett .
b 642 ( 2006 ) 1 [ hep - th/0609069 ] ; m. r. setare , jcap 0701 ( 2007 ) 023 [ hep - th/0701242 ] ; q. wu , y. gong , a. wang , j. s. alcaniz , phys .
b 659 ( 2008 ) 34 [ arxiv : astro - ph/0705.1006 ] ; c. feng , b. wang , y. gong , r. k. su , jcap 0709 ( 2007 ) 005 [ arxiv : astro - ph/0706.4033 ] ; k. karwan , jcap 0805 ( 2008 ) 011 [ arxiv : astro - ph/0801.1755 ] ; r. horvat , phys . rev .
d 70 ( 2004 ) 087301 ; d. pavon , w. zimdahl , phys .
b 628 ( 2005 ) 206 ; b. guberina , r. horvat , h. nikolic , jcap 0701 ( 2007 ) 012 ; x. zhang , f. q. wu , phys .
d 72 ( 2005 ) 043524 ; z. chang , f. q. wu , x. zhang , phys .
b 633 ( 2006 ) 14 ; x. zhang , f. q. wu , phys . rev .
d76 ( 2007 ) 023502 ; x. zhang , int . j. mod .
d 14 ( 2005 ) 1597 ; m. r. setare , j. zhang , x. zhang , jcap 0703 ( 2007 ) 007 ; j. zhang , x. zhang , h. liu , phys .
b659 ( 2008 ) 26 ; x. zhang , phys .
b 648 ( 2007 ) 1 ; x. zhang , phys . rev .
d74 ( 2006 ) 103505 ; j. zhang , x. zhang , h. liu , phys .
b651 ( 2007 ) 84 ; y. ma , x. zhang , phys .
b 661 ( 2008 ) 239 ; j. zhang , x. zhang , h. liu , eur . phys .
j. c 52 ( 2007 ) 693
. m. li , x. d. li , s. wang , y. wang , x. zhang , arxiv:0910.3855 .
s. nesseris , l. perivolaropoulos , phys .
d 72 ( 2005 ) 123519 [ arxiv : astro - ph/0511040 ] ; l. perivolaropoulos , phys .
d 71 ( 2005 ) 063503 [ arxiv : astro - ph/0412308 ] ; s. nesseris , l. perivolaropoulos , jcap 0702 ( 2007 ) 025 [ arxiv : astro - ph/0612653 ] ; e. d. pietro , j. f. claeskens , mon . not .
astron . soc . 341
( 2003 ) 1299 [ arxiv : astro - ph/0207332 ] .
simon , j. , verde , l .
, & jimenez , r. , phys . rev .
d 71 ( 2005 ) 123001 ; gaztaaga , e. , cabr , a. , & hui , l. 2009 , mnras , 399,1663 ; stern , d. , et al .
2010 , jcap 1002 ( 2010 ) 008 ; moresco , m. , et al .
2012 , j. cosmology astropart .
phys . , 1208 , 006 ; busca , n. g. , et al .
, [ arxiv : astro - ph/1211.2616 ] ; omer farooq , data mania , bharat ratra , [ arxiv : astrp - ph/1308.0834 ] .
l. xu , w. li , j. lu , mod .
lett . a 24 ( 2009 ) 1355 [ arxiv:0810.4730 ] ; x. zhang , phys . rev .
d 79 ( 2009 ) 103509 [ arxiv:0901.2262 ] ; c. j. feng x. z. li , phys .
b 680 ( 2009 ) 184 ; m. li , x. d. li , s. wang , x. zhang , jcap 0906 ( 2009 ) 036 [ arxiv : astro - ph/0904.0928 ] .
m. li , c. lin , y. wang , jcap 0805 ( 2008 ) 023 ; q. g. huang , y. g. gong , jcap 0408 ( 2004 ) 006 [ astro - ph/0403590 ] ; x. zhang , f. q. wu , phys .
d 72 ( 2005 ) 043524 [ astro - ph/0506310 ] ; z. chang , f. q. wu , x. zhang , phys .
b 633 ( 2006 ) 14 [ astro - ph/0509531 ] ; x. zhang , f. q. wu , phys . rev .
d 76 ( 2007 ) 023502 [ astro - ph/0701405 ] ; h. zhang , w. zhong , z. h. zhu , s. he , phys . rev .
d 76 ( 2007 ) 123508 [ arxiv : astro/0705.4409 ] ; x. chen , j. liu , y. gong , chin .
lett . 25 ( 2009 )
3086 [ arxiv : gr - qc/0806.2415 ] .
e. elizalde , s. nojiri , s. d. odintsov , p. wang , phys . rev .
d 71 ( 2005 ) 103504 [ hep - th/0502082 ] ; s. nojiri , s. d. odintsov , gen .
38 ( 2006 ) 1285 [ hep - th/0506212 ] ; e. n. saridakis , phys
b 661 ( 2008 ) 335 [ arxiv : gr - qc/0712.3806 ] ; e. n. saridakis , phys . lett .
b 671 ( 2009 ) 331 [ arxiv : gr - qc/0810.0645 ] .
e. ebrahimi , a. sheykhi and h. alavirad arxiv:1209.3147 ; a. banijamali , m.r .
setare , b. fazlpour int .
50 ( 2011 ) 3275 - 3283 ; m.r .
setare phys .
b642 ( 2006 ) 1 - 4 ; q. g. huang and m. li jcap 0408 ( 2004 ) 013 ; a. sheykhi phys .
b680 ( 2009 ) 113 - 117 ; j. zhang , m. zhao , j. lei and x. zhang , eur .
j. c 74 ( 2014 ) 3178 ; b. majeed , m. jamil , a. a. siddiqui , int . j. theor .
doi 10.1007/s10773 - 014 - 2197 - 3 .
k. ichikawa et al jcap 12 ( 2006 ) 005 ; m. jamil , m.u .
farooq , j. cosmol .
astropart . phys .
03 , 001 ( 2010 ) ; m. jamil , e. n. saridakis , j. cosmol .
7 , 28 ( 2010 ) ; m. jamil , a. sheykhi , int
. j. theor .
50 , 625 ( 2011 ) ; m. jamil , e. n. saridakis , setare , m.r . ,
j. cosmol .
phys.11 , 32 ( 2010a ) ; k. karami , a. sorouri , phys .
82 , 025901 ( 2010 ) ; k. karami , et al .
43 , 27 ( 2011a ) ; k. karami , et al . ,
93 , 69001 ( 2011b ) ; k. karami , m. s. khaledian , m. jamil , phys .
83 , 025901 ( 2011c ) ; k. karami , n. sahraei , s. ghaffari , j. high energy phys .
1108 , 150 ( 2011d ) ; k. karami , a. sheykhi , n. sahraei , s. ghaffari , europhys . lett .
93 , 29002 ( 2011e ) ; k. karami , et al .
90 , 473 ( 2012a ) ; k. karami , m. jamil , m. roos , s. ghaffari , a. abdolmaleki , astrophys .
space sci .
340 , 175 ( 2012b ) . | we extend our previous analysis and consider the interacting holographic ricci dark energy ( irde ) model in non - flat universe .
we study astrophysical constraints on this model using the recent observations including the type ia supernovae ( snia ) , the baryon acoustic oscillation ( bao ) , the cosmic microwave background ( cmb ) anisotropy , and the hubble parameter .
it is shown that the allowed parameter range for the fractional energy density of the curvature is @xmath0 @xmath1 @xmath2 @xmath1 @xmath3 in the presence of the interactions between dark energy and matter . without the interaction , the flat universe is observationally disfavored in this model . |
young supernova remnants ( snrs ) are believed to be the prime accelerators of cosmic rays , at least up to the `` knee '' of the cosmic ray spectrum at energies of @xmath2 tev , beyond which the spectrum steepens . however , the direct evidence for particle acceleration to such energies is scarce .
radio emission from shell - type snrs is the result of synchrotron radiation from shock accelerated electrons , but these electrons have typical energies of order 10 gev .
it is only the recent discoveries of nonthermal x - ray emission from shell - type snrs that have finally provided evidence of particles at energies of 10100 tev .
the x - ray emission from sn 1006 ( koyama et al .
1995 ) and [email protected] ( koyama et al . 1997 , slane et al . 1999 ) is _ predominantly _ nonthermal , while that from cas a , kepler , tycho , and rcw 86 ( allen , gotthelf , & petre 1999 ) is predominantly thermal , but contains a nonthermal component as well .
however , it may also be the x - ray emission that poses the most significant constraint on accepting snrs as the source of cosmic rays at energies near the knee .
the x - ray fluxes from all remnants fall below the extrapolation of their radio spectra , implying a steepening or cut - off of the electron spectrum at higher energies .
the implied spectral cut - off energies appear to be well below the knee of the cosmic ray spectrum ( reynolds & keohane 1999 ) .
it is thus of considerable importance to identify new examples of shell - type snrs for which the x - ray emission is largely nonthermal .
rx j0852.0 - 0462 was discovered by aschenbach ( 1998 ) using data from the rosat all - sky survey .
situated along the line of sight to the vela snr , the emission stands out above the soft thermal emission from vela only at energies above @xmath3 kev .
data from the rosat pspc indicates a hard spectrum with either a hot thermal component with @xmath4 kev or a power law component with photon index @xmath5 ( aschenbach 1998 ) .
the remnant is circular with a diameter of @xmath6 ; the emission is particularly enhanced along the northern , western , and southeastern limbs .
hereafter , we use the designation [email protected] for this newly identified snr . as first noted by aschenbach ( 1998 )
the radio emission from [email protected] is quite weak ( @xmath7 jy at 1 ghz ) with faint , limb - brightened emission similar to the x - ray morphology ( combi , romero , & benaglia 1999 ; duncan & green 2000 ) .
sn1006 and g347 are also faint radio emitters , perhaps suggesting a general characteristic of remnants for which the x - ray flux is dominated by nonthermal processes .
iyudin et al .
( 1998 ) reported the comptel detection of @xmath1ti from the source gro j0852 - 4642 , which was tentatively associated with [email protected] .
if correct , this association would be profound ; the very short @xmath1ti lifetime ( @xmath8 y ) would imply a very young snr , and the large observed size would require that the remnant be very nearby as well .
estimates based on the x - ray diameter and @xmath9-ray flux of @xmath1ti indicate an age of @xmath10 y and a distance of @xmath11 pc ( aschenbach , iyudin , & schnfelder 1999 ) .
the hard x - ray spectrum would seem to support this scenario as well .
the inferred temperature would imply a rapid shock that is consistent with a young snr .
however , as we show here , the hard x - ray emission is not from hot , shock - heated gas ; it is nonthermal .
further , separate reanalysis of the comptel data finds that the detection of [email protected] as a @xmath1ti source is only significant at the @xmath12 confidence level ( schnfelder et al .
2000 ) . in the absence of such emission , and given that the x - ray emission is nonthermal , the nearby distance and young age may need to be reexamined .
= 3.5 mm fig .
1. asca gis image of g266.21.2 ( @xmath13 kev ) .
the image consists of a mosaic of 7 individual fields .
contours represent the outline of the vela snr as seen in rosat survey data with the pspc .
we have carried out x - ray observations of [email protected] with the _ advanced satellite for cosmology and astrophysics _ ( asca ) .
the remnant was mapped in 7 distinct pointings , each of @xmath14 ks duration . the resulting image from the asca gis detectors
is illustrated in figure 1 ( see also tsunemi et al .
bright emission along the northern and western shell is accompanied by a compact central source surrounded by diffuse emission .
additional enhanced emission along the southeast shell is seen in the rosat image ( aschenbach 1998 ) but was inadvertently missed in the mapping carried out here .
contours in figure 1 correspond to soft emission ( @xmath15 kev ) from the vela snr , as observed in the rosat all - sky survey .
standard screening processes were applied to data from both the gis and sis detectors , and spectra were extracted from distinct regions of the remnant .
regions outside the snr shell in the northwestern pointing were used for background subtraction for the gis and for sis chips s0c1,2 and s1c0,3 . for the other sis chips , for which no local background was available , we extracted background spectra from the blank sky fields available from the asca guest observer facility .
although these fields are at high galactic latitude , and are thus not representative of our viewing direction ( particularly given the contribution from vela itself ) , we find that results from spectral fitting are quite similar with either background source . in figure 2 we present plots of the x - ray spectra from three regions along the rim of the remnant using the gis detectors : 1 ) the bright northwest rim ; 2 ) the northeast rim ; and 3 ) the western rim .
spectra were extracted from circular regions with radius @xmath16 arcmin centered on the location of maximum brightness . unlike the line - dominated thermal emission that one expects from a young remnant , the spectra are featureless and well described by a power law of index @xmath17 .
this is remarkably similar to that observed for [email protected] ( koyama et al .
1997 , slane et al .
1999 ) , another shell - type snr dominated by nonthermal emission .
the index is flatter than that observed for sn 1006 ( @xmath18 ) , the prototype of this class of nonthermal shell - type snrs ( koyama et al .
1995 ) , although this value was derived for only a portion of the remnant shell .
due to the smaller field of view and the difficulty of merging spectra from different ccds , the spectra from the sis are somewhat sparse .
however , there is still a clear lack of emission line features .
the spectra are still adequately described by an absorbed power law , although the best - fit parameters differ somewhat from those derived from the gis observations . due to the better overall statistics , and the higher sensitivity to the high energy end of the spectrum , we use the results from the gis
spectral fits in the discussion below .
although we have applied a background subtraction to the gis spectra , the soft thermal emission from the vela snr is spatially variable , and there is significant residual soft flux in some of the regions , as seen in figure 2 .
we have modeled the emission with two components . a thermal model ( raymond & smith 1977 ) was used to account for the soft emission , with the column density fixed at @xmath19 , a typical value for the vela snr ( bocchino , maggio , & sciortino 1999 ; aschenbach , egger , & trmper 1985 ) . an absorbed power law was used for the hard emission .
to emphasize the contribution from vela , we have not included the thermal component in the spectral plots shown here . the best - fit spectral parameters for
[email protected] are summarized in table 1 .
the temperature of the thermal component is similar to that observed for the hotter component of vela ( bocchino , maggio , & sciortino 1999 ) , as expected since the gis response is not well - suited to detected the softer component at @xmath20 kev .
the spectral indices for the emission regions along the shell of [email protected] are all consistent with a value of @xmath21 = 3.5 mm fig . 2. asca spectra from both gis detectors for regions of g266.21.2 . the featureless spectrum is well described by a power law . excess flux at low energies is presumably associated with soft thermal emission from the vela snr .
the column density for the power law component summarized in table 1 is significantly higher than that for vela .
while simple scaling of the column density to estimate the distance to [email protected] is clearly rather uncertain , it would appear that the remnant is at least several times more distant than vela . to provide some measure of the scaling of quantities , in discussions below
, we express the distance as @xmath22 .
we note that the column density derived from pspc data alone ( aschenbach 1998 ) , though not well determined , is apparently lower than that measured here . given the importance of the distance measurement , as we discuss below , it is of considerable interest to obtain improved measurements of the column density . + l|llllregion & @xmath23(kev ) & @xmath24 & @xmath25 ( photon ) & @xmath26 +
nw rim & @xmath27 & @xmath28 & @xmath29 & @xmath30 + ne rim & @xmath31 & @xmath32 & @xmath29 & @xmath33 + w rim & @xmath34 & @xmath35 & @xmath36 & @xmath37 + center@xmath38 & @xmath39 & @xmath40 & @xmath41 & @xmath42 + + bb & @xmath43 & @xmath44 & & @xmath45 + pl & & @xmath46 & @xmath47 & @xmath48 + @xmath49 power law component only for snr , in units of @xmath50 + @xmath51 ( 0.5 - 10 kev ) + @xmath52 diffuse emission only as shown in figure 1 , the central region of the remnant contains a compact source which we designate as ax j0851.9 - 4617.4 .
the image centroid is at ra@xmath53 : 08@xmath5451@xmath5557@xmath56 , dec@xmath53 : @xmath5717@xmath5824@xmath59 ; the position uncertainty is roughly 2 arcmin in each direction due to the low - resolution mode in which the gis data were obtained .
this source is surrounded by diffuse emission that extends toward the northwest .
it is of considerable interest to speculate as to whether this represents a central neutron star surrounded by a synchrotron nebula .
such an association would be spectacular indeed , providing further evidence of a massive progenitor for the remnant , and possibly constraining the age through modeling of the plerionic component . in figure 3
we present the gis spectrum for both ax j0851.9 - 4617.4 and the diffuse central region of the remnant .
the diffuse emission is well described by a power law of spectral index @xmath60 , again accompanied by a soft thermal component that we associate with vela . given the nonthermal nature of the shell emission , however , it is quite possible that the diffuse central emission is just associated with emission from the shell projected along the central line of sight .
we note that the spectrum of this central emission appears somewhat harder than that from the rest of the remnant , perhaps suggesting a plerionic nature , but more sensitive observations are required to clarify this .
the gis count rate for ax j0851.9 - 4617.4 is @xmath61 .
the @xmath62 ks gis exposure ( after screening ) thus yields roughly 500 counts in each detector .
the spectrum is thus rather sparse and can be adequately described by a variety of models .
two of particular interest are blackbody emission , as might be expected from the surface of a cooling neutron star or from the polar caps of a neutron star heated through particle acceleration in the magnetosphere , and a power law spectrum that might result from emission directly from a neutron star magnetosphere .
best - fit spectral results from these models are summarized in table 1 .
= 3.5 mm fig .
3. asca spectra from both gis detectors for the central point source and surrounding diffuse region of g266.21.2 .
the diffuse emission is indistinguishable from that of the snr shell .
the point source spectrum is adequately described by a power law , but other models are possible as well .
the prevalence of nonthermal x - ray emission in [email protected] suggests that the remnant is an efficient accelerator of particles .
while thermal emission from swept - up material and shocked ejecta typically dominates the emission of shell - type snrs , there are conditions which can lead to enhanced nonthermal x - ray emission .
recent models for nonlinear shock acceleration in snrs ( e.g. ellison , berezhko , & baring 2000 ) show that the ambient density and magnetic field are critical parameters in determining the relative contributions of thermal and nonthermal electrons in the x - ray emission .
low values of both quantities can yield a synchrotron - dominated remnant in x - rays , while a low magnetic field , in particular , also leads to an increased ratio of the tev @xmath9-ray to radio flux ratio .
this scenario appears consistent with both sn1006 and [email protected] .
the former is the remnant of a type ia explosion , for which the ejecta mass and ambient density are expected to be relatively low , and is located at a relatively high galactic latitude which is consistent with a low magnetic field .
the latter , on the other hand , is in the galactic plane and appears consistent with a massive star progenitor if the association with surrounding molecular clouds and a nearby region suggested by slane et al .
( 1999 ) is correct .
this , too , would then imply a low ambient density if the remnant has evolved in a stellar wind cavity , which is likely in such a scenario .
both remnants are observed in tev @xmath9-rays ( tanimori et al .
1998 , muraishi et al .
2000 ) , and both are weak radio remnants ( stephenson et al . 1977 , aschenbach 1998 , combi et al .
we note , however , that the circumstances which lead to a dominant nonthermal contribution in the soft x - ray band are complicated .
reynolds ( 1998 ) has shown that the maximum nonthermal x - ray emission relative to the radio emission occurs when the remnant age is comparable to the electron loss time in the postshock magnetic field . however
, this alone is not sufficient to guarantee that the thermal x - ray emission from shock - heated ejecta and interstellar material will not dominate the x - ray flux . for [email protected]
, the asca observations clearly show that nonthermal processes dominate the x - ray emission .
our limit on the density of any thermally emitting material in the northwest region , where the emission is brightest , is @xmath63 , where @xmath64 is the filling factor of a sphere taken as the emitting volume in the region extracted .
here we have assumed an equilibrium thermal model ( raymond & smith 1977 ) .
nonequilibrium ionization models can yield higher emissivities , which would lower the derived limit on the density .
we note that this limit is restricted to gas with temperatures above 1 kev because of the severe contributions of soft thermal emission from the vela snr ; higher densities of cooler material in [email protected] can not be ruled out .
observations of the brightest shell regions with high angular resolution , so as to minimize these contributions , are of considerable interest to further constrain the ratio of thermal to nonthermal emission . if [email protected] is actually a source of @xmath1ti at levels indicated by the comptel measurements , so that the age and distance estimates suggested by aschenbach et al .
( 1999 ) and others are correct , then the ambient density is extremely low . the reported @xmath9-ray line flux would indicate a massive star progenitor , and the low density would then imply the presence of a stellar - wind cavity such as that inferred for [email protected] .
= 3.5 mm fig .
4. co emission ( @xmath65 ) along the line of sight to [email protected] .
countours correspond to the x - ray emission from the remnant as seen in figure 1 .
note the use of galactic coordinates to emphasize orientation relative to the galactic plane .
co data ( may , murphy , & thaddeus 1988 ) reveal a concentration of giant molecular clouds the vela molecular ridge at a distance of @xmath66 kpc in the direction of vela .
the presence of ob and r associations , as well as regions , demonstrate that these clouds are sites of recent star formation .
the bulk of the co emission along the line of sight to [email protected] is concentrated in a velocity range @xmath67 ( murphy & may 1991 ) , and extends down to a galactic latitude of @xmath68 .
the total column density through the ridge is in excess of @xmath69 , with the northeast rim of [email protected] ( nearest the galactic plane ) falling along the steeply increasing column density region of the ridge while the western rim lies along a line of much lower column density ( figure 4 ) ; the co column density varies by more than a factor of 6 between these regions .
the lack of a strong variation in @xmath70 across [email protected] indicates that the remnant can not be more distant than the vela molecular ridge . on the other hand ,
if @xmath70 is much larger than for the vela snr , as the gis data reported here indicate , then [email protected] must be as distant as possible consistent with being in front of most of the molecular ridge gas .
this is consistent with an interpretation in which [email protected] is the product of a massive progenitor from this star - forming region , and that the remnant evolution has proceeded in the presence of a wind cavity , leading to a low ambient density much like that indicated for [email protected] .
the angular size of [email protected] gives a free - expansion age of @xmath71 yr , where @xmath72 is the expansion velocity in units of @xmath73 .
alternatively , using the limit on the preshock density derived above , the sedov age of the remnant is @xmath74 yr , where @xmath75 is the kinetic energy from the explosion in units of @xmath76 erg . in either case
, the remnant is relatively young , although ages as young as those based on the @xmath1ti lifetime would require either an extremely large expansion velocity or an extremely low ambient density unless the remnant distance is actually much less than 1 kpc .
we note that the @xmath77 mev @xmath9-rays that announce the presence of @xmath1ti are actually secondary products .
the decay of @xmath1ti proceeds via electron capture to form @xmath1sc which then decays rapidly to @xmath1ca , producing the observed @xmath9-rays .
as electrons fill the vacancy created by the electron capture , we expect sc - k emission of x - rays at @xmath78 kev . using the fluorescence efficiency of @xmath79 ( krause 1979 )
, we expect a flux @xmath80 based on the @xmath1ti flux of iyudin et al .
we have searched the asca sis data for evidence of such a line feature , but have found no compelling evidence to support its presence .
the sis0 spectrum of one region along the nw rim contains a feature at @xmath81 kev ( figure 5 ) , but is not seen in the sis1 spectrum of the same region , nor in adjacent regions with either detector . using our power law fits , the inclusion of the additional feature is not significant according to the f - test ; the thermal fits of tsunemi et al .
( 2000 ) suggest that the feature is significant at the @xmath82 confidence level .
we derive a @xmath83 upper limit of @xmath84 for sc - k emission from this region , which is not inconsistent with the expected result , particularly since additional contributions could come from elsewhere in the remnant .
higher sensitivity searches for such emission are clearly of interest to provide an independent method for obtaining evidence of the presence of @xmath1ti . the nature of the central compact x - ray source in [email protected] is poorly constrained by the asca and rosat data .
there are two bright stars ( @xmath85 ) located within the large positional uncertainty ( aschenbach 1998 ) , but neither is a compelling counterpart .
the first star , wray 16 - 30 ( iras 08502 - 4606 ) , is a be star whose position is within 2 arcmin of ax j0851.9 - 4617.4 . using a thermal model appropriate for
a be star , the x - ray luminosity of the source is @xmath86 . the typical range for
be stars is roughly @xmath87 ( zinnecker & preibisch 1994 ) .
this would suggest a distance of less than @xmath88 pc , which is difficult to reconcile with the observed visual magnitude of @xmath89 the second potential counterpart is hd76060 ( @xmath90 , @xmath91 ) , a b8 iv / v star whose measured parallax puts the distance at @xmath92 pc .
the upper end of the x - ray luminosity for stars of this class is @xmath93 ( berghfer , schmitt , danner , & cassinelli 1997 ) which , for the measured value , requires a distance of roughly 90 pc .
while this discrepancy is perhaps not sufficient to absolutely rule out an association , further observations to improve the position of the x - ray source , and better constrain its spectrum , are clearly in order .
= 3.5 mm fig . 5. asca sis spectrum ( chip s1c0 ) for a region along the nw limb .
there is some evidence of a feature at @xmath94 kev , perhaps suggestive of @xmath1sc emission .
however , this feature is not seen in other sis data .
alternatively , we have considered the possibility that ax j0851.9 - 4617.4 is a neutron star associated with [email protected] . in this case
, the power - law fit yields a spectral index that is rather steep compared with known pulsars .
the luminosity , on the other hand , is quite reasonable for a young neutron star if the distance is indeed several kpc , and the observed column density is compatible with that for the snr shell .
a blackbody model leads to an inferred surface temperature of @xmath95 kev with an emitting region only @xmath96 in radius which could be indicative of emission from compact polar cap regions of a neutron star , although this region is somewhat large for such a scenario .
we have searched for pulsations but detect no obvious modulation of the flux .
we set an upper limit to the pulsed fraction of @xmath97 for frequencies from @xmath98 hz .
finally , we note that the spectrum of the diffuse emission in the central region of the remnant is consistent with a power law whose index is similar to the crab nebula and other synchrotron nebulae associated with supernova remnants .
the ratio of the point source flux to that of the diffuse surrounding emission is @xmath99 which is similar to that for the vela pulsar and its associated synchrotron nebula .
it is of interest that the vela pulsar also has a soft spectrum best characterized as blackbody emission from heated polar caps ( gelman , finley , & zimmermann 1993 ) .
high resolution radio observations of the central region of [email protected] using the atca reveal a faint , slightly extended ( @xmath100 across ) source within the error circle of ax j0851.9 - 4617.4 ( b. gaensler private communication ) .
the 1.4 ghz flux density of this source is @xmath101 mjy , however , substantially lower than seen for known synchrotron nebulae , and the x - ray luminosity is quite low as well .
further observations are required to better understand the nature of this source .
the asca observations of [email protected] reveal that the soft x - ray emission from this snr is dominated by nonthermal processes .
this brings to three the number of snrs in this class , and provides additional evidence for shock acceleration of cosmic rays in snrs .
because of the bright and spatially varying background caused by the vela snr , limits on the thermal emission from [email protected] are difficult to establish
. as a result , its evolutionary state is poorly constrained . however , the asca data reveal a larger column density for this remnant than for vela , indicating that [email protected] is at a larger distance , and perhaps associated with the star formation region located at a distance of @xmath102 kpc . the very strong ratio of nonthermal to thermal x - ray emission argues for a low density environment , possibly suggesting evolution in a stellar wind cavity .
such a scenario would indicate a massive progenitor , from which one could expect a relic neutron star .
the point source ax j0851.9 - 4617.4 could represent this source , with the diffuse emission around the source being an associated synchrotron nebula , albeit a very faint one .
higher resolution x - ray observations of this source are of considerable importance in order to assess its relationship to the snr .
we note that tev @xmath9-ray emission has been detected from both sn 1006 and [email protected] , confirming the presence of very energetic electrons in these remnants .
similar observations of [email protected] are clearly of considerable interest as well .
the authors would like to thank roland diehl and donald ellison for particularly helpful discussions in the preparation of this paper , and also bryan gaensler , douglas bock , and anne green for kindly providing information on radio measurements of ax j0851.9 - 4617.4 , and tom dame for providing the co data .
this work was supported in part by the national aeronautics and space administration through contract nas8 - 39073 and grant nag5 - 9106 , and has made use of data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center .
allen , g. e. , gotthelf , e. v. , petre , r. 1999 , `` evidence of 10 - 100 tev electrons in supernova remnants , '' in proceedings of the 26th international cosmic ray conference , salt lake city , 17 - 25 august 1999 , edited by d. kieda , m. salamon , and b. dingus , vol . 3 , 480 - 3 | the newly discovered supernova remnant [email protected] ( rx j0852.00462 ) , along the line of sight to the vela snr , was observed with asca for 120 ks .
we find that the x - ray spectrum is featureless , and well described by a power law , extending to three the class of shell - type snrs dominated by nonthermal x - ray emission . like [email protected] ,
this low - latitude remnant displays discrete regions of enhanced emission along the rim as well as faint nonthermal emission from the interior .
we derive limits on the thermal content of the remnant emission , although the presence of the vela snr compromises our ability to seriously constrain a low temperature component .
limits placed on the amount of sc - k emission are compared with the expected flux based upon the reported @xmath1ti emission from [email protected] .
we also report on an unresolved x - ray source surrounded by diffuse emission near the center of the remnant .
the properties of the source are not well determined , but appear consistent with the interpretation that the source is a neutron star surrounded by a synchrotron nebula .
alternatively , the source may be associated with one of two stars located within the positional error circle , but this appears somewhat unlikely .
= -0.4truein = -.9truein = 7.3truein = 9.4truein = 10001 |
ever since the seminal work of kramers on the diffusion model of chemical reactions was published about half a century ago @xcite , the theory of activated processes has become a central issue in many areas of science @xcite , notably in chemical physics , nonlinear optics and condensed matter physics .
kramers considered a model brownian particle trapped in a one - dimensional well representing the reactant state which is separated by a barrier of finite height from a deeper well signifying the product state .
the particle was supposed to be immersed in a medium such that the medium exerts a frictional force on the particle but at the same time thermally activates it so that the particle may gain enough energy to cross the barrier . over several decades the model and
many of its variants have served as standard paradigms in various problems of physical and chemical kinetics to understand the rate in multidimensional systems in the overdamped and underdamped limits @xcite , effect of anharmonicities @xcite , rate enhancement by parametric fluctuations @xcite , the role of non - gaussian white noise @xcite , role of a relaxing bath @xcite , quantum and semiclassical corrections @xcite to classical rate and related similar aspects .
the vast body of literature has been the subject of several reviews @xcite and monograph @xcite .
the common feature of overwhelming majority of the aforesaid treatments is that the system is thermodynamically closed which means that the noise of the medium is of _ internal _ origin so that the dissipation and fluctuations get related through the fluctuation - dissipation relation @xcite .
however , in a number of situations the system is thermodynamically open , i.e. , when the system is driven by an _ external _ noise which is _ independent _ of system s characteristic damping @xcite .
the distinctive feature of the dynamics in this case is the absence of any fluctuation - dissipation relation . while in the former case
a zero current steady state situation is characterized by an equilibrium boltzmann distribution , the corresponding situation in the latter case is defined only by a steady state condition , if attainable .
it may therefore be anticipated @xcite that the independence of fluctuations and dissipation tends to make the steady state distribution function depend on the strength and correlation time of external noise as well as on the dissipation of the system .
the elucidation of the role of this steady state distribution in rate theory is worth - pursuing .
our aim in this paper is to generalize kramers theory of activated processes for external noise in this context .
we thus allow the brownian particle in a potential field to be driven by both external and internal stationary and gaussian noise fluctuations with arbitrary decaying correlation functions .
the external noise may be of thermal or non - thermal type .
we consider the stochastic motion to be spatial - diffusion - limited and calculate the rate of escape over the barrier in the intermediate to strong damping regime within an unified description .
the theory we develop here follows closely the original flux over population method of farkas @xcite .
the distinctive aspect , however , is the consideration of a steady state distribution instead of the equilibrium boltzmann distribution for determination of quasi - stationary population in the source well .
this affects the generalized rate expression significantly in two ways .
first , the dynamics around the bottom of the source well exhibits the dependence of steady state distribution on the dissipation .
second , the rate expression remains valid even in absence of any internal thermal noise .
we mention , in passing , that the former point had earlier been rightly emphasized by melnikov @xcite as a specific requirement for a general theory .
some pertinent points regarding the rate theory for nonequilibrium systems may be in order .
it is wellknown that though thermodynamically closed systems with homogenous boundary conditions possess in general time - independent solutions , the driven or open systems may settle down to complicated multiple steady states @xcite when one takes into account of nonlinearity of the system in full .
secondly in most nonequilibrium systems the lack of detailed balance symmetry gives rise to severe problem in determination of stationary probability density for multidimensional problem @xcite . because of its one - dimensional and linearized description the present treatment is free from these difficulties .
it is important to point out that the externally generated nonequilibrium fluctuations can bias the brownian motion of a particle in an anisotropic medium and may used for design of molecular motors and pumps @xcite .
the nonequilibrium , non - thermal systems has also been investigated by a number of worker in different contexts , e.g. , for examining the role of colour noise in stationary probabilities @xcite , the properties of nonlinear systems @xcite , the nature of cross - over @xcite , the effect of monochromatic noise @xcite , the rate of diffusion - limited coagulation processes @xcite , etc . the outlay of the paper is as follows : in sec.ii we generalize kramers theory of reaction rate for external noise .
the stationary , gaussian noise processes are of both external and internal type with arbitrary decaying correlation functions . a general form of steady state distribution function in the source well and a rate expression for barrier crossing dynamics for the nonequilibrium open system have been pointed out . in sec.iii
we explicitly calculate the detailed form of the rate expressions for the specific cases .
the paper is concluded in sec.iv .
we consider the motion of a particle of unit mass moving in a kramers type potential @xmath0 such that it is acted upon by random forces @xmath1 and @xmath2 of both internal and external origin , respectively , in terms of the following generalized langevin equation @xmath3 where the friction kernel @xmath4 is connected to internal noise @xmath1 by the wellknown fluctuation - dissipation relationship @xcite @xmath5 we assume that both the noises @xmath1 and @xmath2 are stationary and gaussian .
their correlation times may be of arbitrary decaying type .
the external noise is independent of the memory kernel and there is no corresponding fluctuation - dissipation relation .
we further assume , without any loss of generality , that @xmath1 is independent of @xmath2 so that we have @xmath6 the external noise modifies the dynamics of activation in two ways .
first , it influences the dynamics in the region around the barrier top so that the effective stationary flux across it gets modified .
second , in presence of this noise the equilibrium distribution of the source well is disturbed so that one has to consider a new stationary distribution , if any , instead of the standard boltzmann distribution .
this new stationary distribution must be a solution of the generalized fokker - planck equation around the bottom of the source well region and serve as an appropriate boundary condition analogous to kramers problem .
we consider these two aspects separately in the next two subsections .
we consider the potential @xmath0 as shown in fig.1 . linearizing the potential around the barrier top at @xmath7 we write @xmath8
thus the langevin equation takes the following form @xmath9 where @xmath10 the general solution of eq.([eq5 ] ) is given by , @xmath11 where @xmath12 with @xmath13 and @xmath14 being the initial position and velocity of the brownian particle that are assumed to be nonrandom , and @xmath15 the kernel @xmath16 is the laplace inversion of , @xmath17 with @xmath18 the time derivative of eq.([eq7 ] ) gives @xmath19 with @xmath20 and @xmath21 now using the symmetry of the correlation function , @xmath22 we compute the explicit expressions of the variances in terms of @xmath16 and @xmath23 as , @xmath24 ^ 2 \rangle \nonumber \\ & = & 2 \int_0^t m_b(t_1 ) \ ; dt_1 \int_0^{t_1 } m_b(t_2 ) \ ; c(t_1 - t_2 ) \ ; dt_2 \ ; \ ; , \\
\sigma_{vv}^2 ( t ) & \equiv & \langle [ v(t ) - \langle v(t ) \rangle ] ^2 \rangle \nonumber \\ & = & 2 \int_0^t m_b(t_1 ) \ ; dt_1 \int_0^{t_1 } m_b(t_2 ) \ ; c(t_1 - t_2 ) \ ; dt_2 \ ; \ ; , \\
\sigma_{xv}^2 ( t ) & \equiv & \langle [ x(t ) - \langle x(t ) \rangle ] \ ; [ v(t ) - \langle v(t ) \rangle ] \rangle \nonumber \\ & = & \int_0^t m_b(t_1 ) \ ; dt_1 \int_0^t
m_b(t_2 ) \ ; c(t_1 - t_2 ) \ ; dt_2\end{aligned}\ ] ] and from ( 14a ) and ( 14c ) we see that @xmath25 while calculating the variances it should be remembered that by virtue of eq.([eq6 ] ) @xmath26 since , in principle we know all the average quantities and variances of the linear system driven by gaussian noise one can make use of the characteristic function method to write down the fokker - planck equation for phase space distribution function @xmath27 near the barrier top @xcite @xmath28 p(x , v , t ) & = & \bar{\gamma}_b ( t ) \frac{\partial}{\partial v } v p(x , v , t ) \nonumber \\ & & + \phi_b(t ) \frac{\partial^2}{\partial v^2 } p(x , v , t ) + \psi_b(t ) \frac{\partial^2}{\partial v \partial x } p(x , v , t)\end{aligned}\ ] ] with @xmath29 regarding the fokker - planck equation ( [ eq16 ] ) three points are to be noted . first , although bounded the time dependent functions @xmath30 , @xmath31 and @xmath32 may not always provide long time limits .
these play a decisive role in the calculation of non - markovian kramers rate .
therefore , in general , one has to work out frequency @xmath33 and friction @xmath34 functions for analytically tractable models @xcite .
second , when the noise is purely internal ( i.e. , there exist a fluctuation - dissipation relation ) we have @xcite @xmath35 \ ; \ ; .\ ] ] third , for pure external noise with markovian relaxation , i.e. , @xmath36 we have @xmath37 in order to calculate the stationary distribution near the bottom of the left well we now linearize the potential @xmath0 around @xmath38 .
the corresponding fokker - planck equation can be constructed using the above - mentioned technique to obtain @xmath39 p(x , v , t ) & = & \bar{\gamma}_0 ( t ) \frac{\partial}{\partial v } v p(x , v , t ) \nonumber \\ & & + \phi_0(t ) \frac{\partial^2}{\partial v^2 } p(x , v , t ) + \psi_0 ( t ) \frac{\partial^2}{\partial v \partial x } p(x , v , t)\end{aligned}\ ] ] with @xmath40 here the subscripts ` 0 ' signifies the dynamical quantities corresponding to the bottom of the left well .
it may be easily checked that the stationary solution of eq.([eq20 ] ) is given by @xmath41\ ] ] where , @xmath42 ; @xmath43 , @xmath44 and @xmath45 are the values at long time limit and @xmath46 is the normalization constant . here
@xmath47 is the renormalized linearized potential with a renormalization in its frequency .
it must be emphasized that the distribution ( [ eq22 ] ) is not an equilibrium distribution .
this stationary distribution for the open system plays the role of an equilibrium distribution for the closed system which may be however recovered in the absence of external noise terms .
we also point out in passing that because of the linearized potential @xmath48 the steady state is unique and the question of multiple steady states does not arise . in the spirit of kramers celebrated ansatz @xcite we now demand a solution of the eq.([eq16 ] ) at the stationary limit of the type @xmath49 \ ; \xi ( x , v)\ ] ] with @xmath50 and @xmath51 are the long time limits of the corresponding time dependent quantities specific for the barrier top region .
the notable difference from the kramers ansatz is that the exponential factor in ( [ eq23 ] ) is not the boltzmann factor but pertains to the dynamics at the barrier top .
the ansatz of the form ( [ eq23 ] ) denoting the steady state distribution is motivated by the local analysis near the bottom and top of the barrier in the kramers sense . for a nonequilibrium system , as in the present problem of external time - dependent potential field ,
the relative population of the two regions , in general , depends on the global properties of the potential .
thus although at equilibrium the probability density is given by a boltzmann distribution , the external modulation of the potential requires energy input and drives the system away from equilibrium , disturbing the boltzmann distribution . at this point
one may anticipate the signature of dynamics in the kramers-like ansatz ( [ eq23 ] ) compared to the standard kramers ansatz for closed system ( i.e. , when the external field is absent ) .
thus while in the latter case one considers a complete factorization of the equilibrium part ( boltzmann ) and the dynamical part , @xmath52 , the ansatz ( [ eq23 ] ) incorporates the additional dynamical contribution through dissipation and the strength of the noise into the exponential part .
this explicit dynamical modification of kramers ansatz in the form of ( [ eq23 ] ) is valid so long as the extra dynamical contribution in the exponential factor in ( [ eq23 ] ) does not become too severe , i.e. , the amplitude of the external noise field is not too strong . to put it in a more quantitative way ,
this implies ( assuming for simplicity @xmath53 , @xmath54 ) that the thermal length scale , i.e. , the maximum value of @xmath55 on which the velocity of the particle is thermalized , should be shorter than the other characteristic length scales of the system , e.g. , @xmath56 these considerations are necessary for making spatial diffusion regime and quasi - stationary condition meaningful in the present context . now inserting ( [ eq23 ] ) in ( [ eq16 ] ) in the steady state
we get @xmath57 \frac{\partial \xi}{\partial v } + \phi_b \frac{\partial^2 \xi}{\partial v^2 } + \psi_b \frac{\partial^2 \xi}{\partial v \partial x } = 0 \ ; \ ; .\ ] ] at this point we set @xmath58 and with the help of the transformation ( [ eq25 ] ) , eq.([eq24 ] ) is reduced to the following form @xmath59 \frac{d \xi}{d u } = 0 \ ; \ ; .\ ] ] now , let @xmath60 where @xmath61 is a constant to be determined later . from ( [ eq25 ] ) and ( [ eq27 ] ) we have @xmath62 with @xmath63 by virtue of the relation ( [ eq27 ] ) , eq.([eq26 ] )
becomes @xmath64 where @xmath65 the general solution of the homogenous differential equation ( [ eq30 ] ) is @xmath66 where @xmath67 and @xmath68 are the constants of integration .
the integral in the eq.([eq32 ] ) converges for @xmath69 if only @xmath70 is positive .
the positivity of @xmath70 depends on the sign of @xmath71 ; so by virtue of eqs.([eq25 ] ) and ( [ eq27 ] ) we find that the negative root of @xmath71 , i.e. , @xmath72 guarantees the positivity of @xmath70 since @xmath73 to determine the value of @xmath67 and @xmath68 we impose the first boundary condition on @xmath74 @xmath75 this condition yields @xmath76 inserting ( [ eq35 ] ) into ( [ eq32 ] ) we have as usual @xmath77 \ ; \ ; .\ ] ] since we are to calculate the current around the barrier top , we expand the renormalized potential @xmath47 around @xmath78 @xmath79 thus with the help of ( [ eq36 ] ) and ( [ eq37 ] ) , eq.([eq23 ] ) becomes @xmath80\ ] ] with @xmath81 now defining the steady state current @xmath82 across the barrier by @xmath83 we have using eq.([eq38 ] ) @xmath84 \ ; \ ; .\ ] ] having obtained the steady state current over the barrier top we now look for the value of the undetermined constant @xmath68 in eq.([eq41 ] ) in terms of the population in the left well .
we show that this may be done by matching two appropriate _ reduced _ probability distributions at the bottom of the left well .
to do so we return to the eq.([eq23 ] ) which describes the steady state distribution at the barrier top . again with the help of ( [ eq36 ] ) we have @xmath85 \ ; \exp \left [ -\frac{v^2}{2d_b } - \frac { \tilde{v}(x)}{d_b+\psi_b } \right ] \ ; \ ; .\ ] ] we first note that , as @xmath86 , the pre - exponential factor in @xmath87 reduces to the following form @xmath88 = f_2 \ ; \left ( \frac{2\pi}{\lambda } \right ) ^{1/2 } \ ; \ ; .\ ] ] we now define a reduced distribution function in @xmath89 @xmath90 hence we have from ( [ eq44 ] ) and ( [ eq45 ] ) @xmath91 \ ; \ ; .\ ] ] similarly we derive the reduced distribution function in the left well , around @xmath92 using ( [ eq22 ] ) as @xmath93\ ] ] where we have employed , the expansion of @xmath48 as @xmath94 and @xmath46 as the normalization constant . at this juncture
we impose the second boundary condition that , at @xmath95 the reduced distribution function ( [ eq46 ] ) must go over to stationary reduced distribution function ( [ eq47 ] ) at the bottom of the left well .
thus we have @xmath96 the above condition is used to determine the undetermined constant @xmath68 in terms of the normalization constant @xmath46 of eq.([eq22 ] ) @xmath97 \ ; \exp \left [ \frac { \tilde{v}(0 ) -\frac{1}{2}\bar{\omega}_b^2 x_a^2 } { d_0+\psi_0 } \right ] \ ; \ ; .\ ] ] evaluating the normalization constant by explicitly using the relation @xmath98 and then inserting its value in ( [ eq50 ] ) we obtain @xmath99 \ ; \ ; .\ ] ] making use of the relation @xmath100 in ( [ eq52 ] ) and then the value of @xmath68 in eq.([eq41 ] ) we arrive at the expression for the normalized current or barrier crossing rate @xmath101 \ ; \ ; .\ ] ] where the activation energy @xmath102 is defined as @xmath103 as shown in fig.1 . since the temperature due to internal thermal noise , the strength of the external noise and damping constant are buried in the parameters @xmath104 , @xmath105 , @xmath43 , @xmath51 and @xmath70 the general expression ( [ eq53 ] ) looks somewhat cumbersome .
we note that the subscripts ` 0 ' and ` b ' in @xmath106 or @xmath107 refer to the well or the barrier top region , respectively .
we discuss it in greater detail in the next section .
eq.([eq53 ] ) is the central result of this paper .
this generalizes the kramers expression for rate of the activated processes for the nonequilibrium open systems .
both the internal and the external noises may be of arbitrary long correlation time .
it is important to note that the pre - exponential dynamical factors as well as the exponential factor are modified due to the openness of the system .
the modification of the exponential factor is due to the fact that depending on the strength of the external noise @xmath2 the system settles down to a stationary distribution which does not coincide with the usual equilibrium boltzmann distribution .
the system therefore attains the steady state at a different ` effective ' temperature .
this aspect will be clarified in greater detail when we consider the limiting case in subsection d. in general , both the factors in the rate depend on the strength of the noise , correlation time of fluctuations of both external and internal noise processes and dissipation .
the rate is spatial - diffusion - limited and is valid for intermediate to strong damping regime .
this validity must be appreciated in the present context of driven system in the sense that while on the one hand thermal length scale of the system must be short compared to other characteristic length scales of the sytem corresponding to the inequality ( [ eq71 ] ) , dissipation should also obey the restriction that during one round trip of the particle in phase space ( in action , angle space ) under purely deterministic motion corresponding to ( [ eq1 ] ) , the energy dissipated is greater than the thermal energy , i.e. , @xmath108 where @xmath109 , the action , is equivalent to unperturbed energy @xmath102 in the weak friction limit .
both the inequalities ( [ eq71 ] ) and ( [ eq72 ] ) are therefore relevant for quantifying the spatial - diffusion - limited intermediate to strong damping regime . in what follows we
shall be concerned with several limiting situations to illustrate the general result ( [ eq53 ] ) systematically for both thermal and non - thermal activated processes .
we first consider the case with no external noise and the internal thermal noise is @xmath110-correlated . to this end
we set @xmath111 making use of the abbreviations in eqs.(17 ) and ( 21 ) it follows after some algebra that @xmath112 the above relations reduce the general expression ( [ eq53 ] ) to classical expression for kramers rate @xcite @xmath113 \ ; e^{-e / k_bt } \ ; \ ; .\ ] ] next we consider the case with no external noise but the internal noise is of ornstein - uhlenbeck type @xcite .
thus we have @xmath114 here @xmath115 denotes the strength while @xmath116 refers to the correlation time of the noise . again from eqs.(17 ) , ( 18 ) and ( 21 ) along with ( [ eq56 ] ) we derive the following relations @xmath117 \ ; \ ; { \rm and } \\
a_\pm = \frac{1}{1+d_b } \ ; \left
[ -\frac { \bar{\gamma}_b } { 2 } \pm \sqrt { \frac { \bar{\gamma}_b^2}{4 } + \bar{\omega}_b^2 } \right ] \ ; \ ; .\end{aligned}\ ] ] and hence the rate becomes @xmath118 \ ;
e^{-e / k_bt } \ ; \ ; .\ ] ] whereby we recover the result of grote - hynes @xcite and hnggi - mojtabai @xcite obtained several years ago .
next we consider the case where the noise is completely due to of external source and the external noise is of ornstein - uhlenbeck type @xcite so that we set @xmath119 note that since in this case the dissipation is independent of fluctuations we may assume markovian relaxation so that @xmath36 ( see also eqs.(18 ) and ( 19 ) ) . the above condition ( [ eq58 ] ) when used in eqs.(17 ) , ( 19 ) and ( 21 ) we obtain after some lengthy algebra @xmath120 \ ; \ ; { \rm and } \\ a_\pm = \frac{1}{1+\gamma \tau_c } \ ; \left [ - \frac{\gamma}{2 } \pm \sqrt { \frac{\gamma^2}{4 } + \omega_b^2 } \right ] \ ; \ ; .\end{aligned}\ ] ] and the rate becomes @xmath121 \ ; \exp \left [ - \frac { \gamma ( 1+\gamma \tau_c - \omega_b^2 \tau_c^2 ) } { { \cal d } ( 1 + \gamma \tau_c ) } e \right ] \ ; .\ ] ] it is interesting to note that the expression ( [ eq59 ] ) denotes the external noise - induced barrier crossing rate which crucially depends on the strength @xmath115 and correlation time @xmath116 of the coloured noise .
the absence of temperature and the appearance of dissipation @xmath122 explicitly demonstrates the non - thermal origin of the noise processes as well as the absence of fluctuation - dissipation relation .
we finally consider both the internal and external noise to be @xmath110-correlated , i.e. , @xmath123 @xmath124 being the strength of the external white noise .
hence , by virtue of ( 15 ) , ( 17 ) and ( 21 ) we have @xmath125 hence the rate becomes @xmath126 \ ; \exp \left [ \frac{e } { k_bt + ( \alpha/\gamma ) } \right ] \ ; \ ; .\ ] ] in the limit @xmath127 we recover the kramers original result ( [ eq55 ] ) for pure internal white noise .
we note here that @xmath128 defines a new ` effective ' temperature due to external noise .
the effective temperature which depends on the strength of the external noise had been discussed earlier by bravo et .
@xcite in a somewhat different context .
we note that while in the latter case the bath is driven by external fluctuations , the present treatment concerns the direct driving of the reaction co - ordinate by external noise .
in this paper we have generalized kramers theory of activated processes for nonequilibrium open systems .
the theory takes into account of both internal and external gaussian noise fluctuations with arbitrary decaying correlation functions in an unified way .
the treatment is valid for intermediate to strong damping regime for spatial diffusive processes .
\(i ) we have shown that not only the motion at the barrier top is influenced by the dynamics , it has an important role to play in establishing the stationary state near the bottom of the source well for the open systems .
thus the stationary distribution function in the well depends crucially on the correlation time of the external noise processes as well as on damping .
this is distinctly a different situation ( but analogous ) as compared to an equilibrium boltzmann distribution in the source well for standard kramers theory for closed systems .
\(iii ) we have checked and examined the various limits of the generalized rate expression to obtain kramers rate , its non - markovian counterpart as well as the other cases for specific external noise processes in presence and absence of the internal noise .
we conclude by noting that since the validity of the rate expression derived in the paper depends on the existence of long time limit of the moments for the stochastic processes , the theory can not be directly extended to , say , fractal noise processes .
these and the related noise processes remain outside the scope of the present treatment .
suitable extension of the kramers theory in this direction is worth - pursuing .
j. ray chaudhuri , g. gangopadhyay and d. s. ray , j. chem . phys . * 109 * , 5565 ( 1998 ) ; j. r. chaudhuri , b. deb , g. gangopadhyay and d. s. ray , j. phys .
b * 31 * , 3859 ( 1998 ) ; j. r. chaudhuri , s. k. banik , b. deb and d. s. ray , eur .
j. d * 6 * , 415 ( 1999 ) . | the kramers theory of activated processes is generalized for nonequilibrium open one - dimensional systems .
we consider both the internal noise due to thermal bath and the external noise which are stationary , gaussian and are characterized by arbitrary decaying correlation functions .
we stress the role of a nonequilibrium stationary state distribution for this open system which is reminiscent of an equilibrium boltzmann distribution in calculation of rate .
the generalized rate expression we derive here reduces to the specific limiting cases pertaining to the closed and open systems for thermal and non - thermal steady state activation processes . |
in the modern viewpoint on gauge symmetry enhancement in open string theory , the gauge vectors associated to a @xmath2 gauge algebra are seen to arise from open strings beginning and ending on a collection of @xmath3 dirichlet branes ( see @xcite ) . given the utility and simplicity of brane pictures for gauge theory it is of interest to understand the brane picture of other gauge algebras .
a natural laboratory for this problem is iib superstrings compactified on a two - sphere in the presence of several 7-branes appearing as points on this sphere and extending along the other spatial dimensions . in this background
both strings and seven branes carry @xmath1 labels . here
@xmath4 and @xmath5 are relatively prime integers denoting for the case of a string the charges under the ns - ns and rr antisymmetric tensors respectively .
a @xmath6 $ ] seven - brane is a seven - brane where a @xmath7 string can end .
the f - theory picture @xcite of this superstring background was used to obtain the specific 7-brane content that would lead to the @xmath8 ( @xmath9 ) and @xmath10 ( @xmath11 ) gauge algebras @xcite .
such gauge algebras require branes of various @xmath6 $ ] types .
it was naturally assumed that open strings , represented by geodesic lines on the two sphere joining suitable branes , would give rise to the gauge vectors of the corresponding algebra .
this is not always the case ; depending on the positions of the branes , a given gauge vector is realized sometimes as a geodesic string , but more generally it is realized as a geodesic string junction @xcite .
indeed , such transitions were examined explicitly in @xcite where strong evidence was given to the effect that given a specific configuration of branes either a unique open string or a unique string junction is geodesic .
string junctions , originally introduced in @xcite , have been the subject of much recent study with regards to their bps properties @xcite , their dynamics @xcite , their role in @xmath12 super - yang - mills @xcite , as well as for their relevance to dualities @xcite and to the analysis of bps states of @xmath13 sym theory @xcite .
there is also a parallel line of development dealing with @xmath6 $ ] five - branes and five dimensional gauge theory @xcite .
the main motivation for considering junctions in @xcite was that such string states can carry manifestly the gauge charges under the obvious subalgebras and this enables one to reproduce the structure of the exceptional algebras by combining junctions the same way one combines open strings .
the purpose of the present paper is to give a firm lie - algebraic basis to junctions , explaining the reason for much of the coincidences noted in @xcite .
while that work dealt with adjoint representations only , we will be able to discuss here arbitrary representations of the _ ade _ series of algebras .
since junctions represent states with gauge charges , they are naturally associated to weight vectors belonging to some representation of the relevant algebra .
the main goal of the present paper is to show how to calculate the weight vector associated with an arbitrary junction , and conversely , given an arbitrary representation of an _ ade _ algebra , to show how to construct the junctions representing the corresponding string states .
it is clear that representations other than adjoints are present in applications , typically as massive bps states @xcite .
some of our results related to @xmath14 symmetry were anticipated in an interesting paper by imamura @xcite , who among other things showed how to construct the @xmath15 and @xmath16 representations .
our treatment will focus only on the lie - algebraic issues , and we will not discuss whether the general junctions we consider have bps representatives .
this is an important issue and should be studied , possibly along the lines of mikhailov _ et al .
_ @xcite .
junctions fall into equivalence classes .
different members of an equivalence class represent the same physical state in different regions of the moduli space of backgrounds .
the different members differ by transformations that involve strings crossing branes , creating or destroying prongs in this process @xcite .
this process is u - dual to the original version of the hanany - witten effect @xcite .
in addition , junctions that differ by smooth deformations are also to be considered equivalent , including the possibility of collapsing or resolving vertices
. since junctions with very different presentations can be equivalent , one must first learn how to tell when junctions are equivalent .
we solve this problem by introducing a set of charges @xmath17 , one for each 7-brane of the background , invariant under the operations that preserve the equivalence class of a junction .
these charges can be used to construct a canonical presentation for any junction , making it manifest that two junctions are equivalent _ if and only if _ their invariant charges agree . for each brane ,
the invariant charge arises from two sources , prongs ending on the brane and strings crossing the branch cut emerging from the brane .
a junction may be localized around the set of seven - branes , or may have an asymptotic string ( or strings ) carrying @xmath1 charge away from the set of branes .
the invariant charges allow us to think of junctions as linear combinations with integer coefficients of a set of basis strings @xmath18 . for each brane ,
the associated basis string is a string departing the brane and going off to infinity without crossing any branch cut . in this way
the space of ( equivalence classes of ) junctions naturally becomes a _ lattice _ over the set of basis strings .
this a very important concept since it immediately gives a complete classification of all possible junctions , and helps make systematic any search for special junctions .
it is important to emphasize that in each case the number of basis strings ( or branes ) exceeds the rank of the lie algebra to be generated by the background .
this means that the basis of strings is not equivalent to a basis of junctions for the roots of the lie algebra . on the other hand ,
the sublattice of junctions with zero asymptotic charge is in fact spanned by a number of junctions equal to the rank .
motivated by the intersection invariant of homology two cycles in the f - theory ( or m - theory ) picture we introduce a symmetric bilinear form @xmath19 or metric on the junction lattice .
this is a slight modification of the standard concept , since our basis strings are not homology cycles . with a little abuse of language , given a junction @xmath20
we call @xmath21 the self - intersection number of the junction .
super - yang - mills in ref . @xcite . ]
the metric is negative - definite for the @xmath22 series .
it is degenerate for the @xmath8 series ; for the @xmath23 algebra there exist two null junctions ( junctions whose intersection with any junction vanish ) while for @xmath8 algebras with @xmath24 there is one null junction .
the metric is non - degenerate but of indefinite signature for the @xmath10 series ; it has @xmath3 negative eigenvalues and two positive ones .
we search the junction lattices for all inequivalent junctions with zero asymptotic charge and self - intersection @xmath25 .
for each case , we find a set that is in correspondence with the set of roots of the corresponding lie algebra .
we suspect that this ( simple ! ) derivation of the lie algebra from the brane configuration must somehow parallel the work involved in understanding the detailed nature of the kodaira singularities .
all roots are presented as junctions , but , completing the work of @xcite we show that they all have equivalent presentations as open strings without self - crossings ( jordan open strings ) .
indeed any jordan open string with endpoints at two different branes has self - intersection @xmath25 , but we are not aware of an argument that would imply that any junction of self - intersection @xmath25 must have a presentation as a jordan open string joining two different branes .
it is therefore interesting to confirm that all roots of the @xmath0 algebras have jordan open string presentations .
we believe this analysis improves substantially our understanding of the appearance of the adjoint representations . as usual , selecting a base of simple roots involves choices .
we make a choice and exhibit junctions @xmath26 representing the simple roots of the @xmath0 algebras , confirming that their intersection numbers give rise to the appropriate cartan matrices .
we define fundamental weight junctions @xmath27 dual to the simple root junctions @xmath26 , and introduce additional junctions with nonzero asymptotic charge to span the entire junction lattice .
the new junctions can be thought of as additional fundamental weights @xmath28 , and allow us to develop an extended formalism .
when the metric in the lattice of junctions is non - degenerate one can define dual roots @xmath29 , and the resulting extended cartan matrix is seen to be block diagonal .
an arbitrary junction can be expanded using the extended set of fundamental weights ; the coefficients of the @xmath28 give the asymptotic charges , and the coefficients of the @xmath27 , we claim , are simply the dynkin labels of the associated weight vector .
proving this requires further analysis of the junction lattice and its relation to the weight lattice .
note that the claim implies that shifting a junction by an @xmath28 does not change its associated weight vector .
we search systematically the lattices for all string junctions of self - intersection @xmath30 , which we expect to be related to fundamental representations .
any jordan open string with one endpoint on a brane and the other at infinity represents a junction of self - intersection @xmath30 .
again , while it may not be guaranteed , for all cases we have examined the junctions of self - intersection @xmath30 have jordan open string representatives .
we then turn to the subject of obtaining general representations .
even though we already have junctions @xmath27 for weights , these are not proper junctions ( except for @xmath31 ) as they require basis strings with fractional coefficients .
thus , given a weight vector @xmath32 one is naively led to a junction @xmath33 which is not proper . here is where the junctions @xmath28 carrying asymptotic charges play a crucial role .
they themselves are not proper , but can be added to the naive weight junction to give a proper junction representing the particular weight .
more precisely , given a specific representation , one can find asymptotic charges @xmath1 such that all weights in the representation are represented by proper junctions .
in addition , we show that for a given algebra , every representation in each conjugacy class can be represented by proper junctions using any one of a list of possible @xmath1 values .
the @xmath1 sets associated to each conjugacy class are disjoint and together they are complete .
moreover , they define an additive group ; given two sets , any sum of charges involving one element of each set falls into a unique third set . finally , the map associating conjugacy classes to @xmath1 sets is an isomorphism between the group of conjugacy classes under tensor product of representations , and the additive group of @xmath1 sets .
thus we elucidate the relation between the junction lattices and the corresponding lie - algebra weight lattice .
the above understanding is obtained by explicit examination of particular examples , followed by generalization .
we aim to relate junction and weight lattices via a relation of the form @xmath34 , with @xmath35 the dynkin labels defining the weight vector associated to the junction with invariant charges @xmath17 .
the goal here is finding the matrix @xmath36 .
we give a detailed discussion for the case of the @xmath37 algebras and the case of the @xmath14 algebra .
the matrix @xmath36 is partially fixed by the constraints of the adjoint representation . for the case of @xmath37
the relation is completely fixed by further consideration of the fundamental representation , whose states we expect to be represented by asymptotic strings ending on the branes . in the case of @xmath14
we see that the junctions of self - intersection @xmath30 fit into the @xmath38 , @xmath39 and @xmath16 representations in a unique fashion , and this fixes the matrix @xmath36 completely . with these examples understood in detail
we notice that in fact , @xmath36 has a simple interpretation : @xmath40 is simply the intersection number between the basis string @xmath18 and the junction @xmath26 representing the root @xmath41 .
this allows a nice compact expression for the dynkin labels of any junction @xmath20 : one readily finds @xmath42 .
this confirms the prior claim that for a given junction , the expansion coefficients of the @xmath27 are the dynkin labels of the associated weight vector .
this expression for @xmath36 is uniquely fixed by the constraint that it gives the correct assignment for the junctions belonging to the adjoint , and by the constraint that two sets of junctions with different nonvanishing asymptotic charges are correctly matched to two representations .
this understanding enables us to deal quite easily with the algebras of the @xmath43 series .
we also consider the reverse problem , given some dynkin labels specifying a weight vector , how do we build the corresponding junction ? here
the key is that we have two extra equations .
along with the dynkin labels we think of the asymptotic charges @xmath4 and @xmath5 as two more labels .
the dynkin labels of the extended framework then match the number of invariant charges .
the equations can then be inverted to give the invariant charges in terms of the extended dynkin labels . at this stage
one can see explicitly how suitable values of @xmath4 and @xmath5 are required to achieve integral invariant charges and the connection to conjugacy classes of representations becomes apparent . before summarizing the organization of this paper we touch on a couple of issues that were somewhat puzzling in @xcite .
we wondered at that time what was the precise lie - algebraic counterpart of the combining of junctions .
the answer is now clear , combination of junctions is simply addition of the corresponding lattice vectors , and therefore equivalent to addition of the corresponding weight vectors .
it is also clear that while junctions associated to roots can always be combined , the result need not be a root .
another point was the following : for the @xmath44 algebra the brane arrangement involved @xmath3 @xmath45-branes , one @xmath46-brane and two @xmath47-branes . on the other hand
the maximal subalgebra of @xmath44 of a similar type is @xmath48 .
this is a little surprising since given the number and types of branes present one would have naively expected a @xmath49 algebra . thus out of three @xmath50 s two are not realized .
those @xmath50 charges are two linear combinations of invariant charges under which no @xmath44 root can be charged .
it is clear that these must simply be the @xmath4 and @xmath5 charges , expressed in terms of the invariant charges .
we wondered in @xcite how one could read the surviving @xmath50 charge from the picture of the junction .
the answer follows from our results , and for convenience we discuss it in an appendix . given a junction ,
we have already shown how to find its dynkin labels for @xmath44 , it then takes a routine lie - algebraic computation to derive the dynkin weights of subalgebras . in this way
one can tell precisely which combination of invariant charges gives the @xmath50 charge in the @xmath48 subalgebra .
this paper is organized as follows . in section 2
we define invariant charges , introduce the canonical presentation of junctions and the basis of strings . in section 3 we introduce a symmetric bilinear intersection form and discuss self - intersection numbers for junctions . after reviewing briefly some basics of representation theory
, we prove that the inequivalent junctions with zero asymptotic charge and self - intersection @xmath25 are in correspondance with the roots of the enhanced symmetry algebra . in section 4
we introduce fundamental weight junctions and the extended formalism . in section 5
we search for all junctions of self - intersection @xmath30 . in section 6
we show how to relate dynkin labels to invariant charges by discussing explicitly @xmath37 , @xmath14 and @xmath9 . in section 7
we establish the relation between the junction and weight lattices , and discuss the @xmath43 series .
finally in section 8 we offer some concluding remarks .
in this section we discuss the equivalence classes of junctions .
two junctions are considered equivalent if they can be related by _
junction transformations_. these transformations are of two types : moving pieces of the junction across branes , and continuous deformations of the junction consistent with charge conservation .
the operation of brane crossing , the presently relevant version of the hanany - witten effect @xcite , can create or destroy string prongs @xcite .
the continuous deformations of a junction include the possibility of coalescing junction points ( see also @xcite ) . our strategy for the classification of junctions will be the following .
we will recognize a set of invariants , the invariant charges , associated to a junction .
we will then explain how any string junction can be given a _
canonical presentation _ determined uniquely by , and in one to one correspondance with the invariant charges .
this will prove that the invariant charges characterize completely the equivalence classes of junctions .
we will be led to define a basis of junctions comprised by strings , and general junctions will be viewed as points in a lattice whose generating vectors are the basis strings .
the invariants associated to a junction will be a set of charges , one for each 7-brane in the background .
our intuition about charges is that a junction may be charged under the gauge field associated to a particular brane either if it has a prong on the brane , or if it crosses the branch cut associated to the brane . for each brane ,
the invariant is simply the charge which combines the two possible contributions so that the charge is unchanged when the junction is modified by crossing transformations .
such charges exist because crossing transformations are indeed constrained by charge conservation @xcite . consider a junction @xmath51 and let @xmath52 index the various branes in the configuration .
let @xmath53 $ ] denote the labels of a brane , and let @xmath54 denote the branch cut emerging from the brane and oriented away from it .
we now claim that the junction has an invariant charge @xmath55 associated to the @xmath52 brane given by @xmath56 in the above expression , @xmath57 is the number of @xmath58 prongs departing from the brane , and @xmath59 is the number of @xmath58 prongs ending on the brane .
moreover , @xmath60 , an integer greater than or equal to zero , denotes the number of intersections of @xmath20 with the branch cut @xmath61 . finally , @xmath62 is the charges of the string belonging to @xmath20 that crosses the cut at the @xmath63-th intersection point , in a counterclockwise direction .
@xmath64 let us now confirm this is the correct formula .
the @xmath65 term in eqn .
( [ invch ] ) gives the net number of outgoing prongs at the brane .
in addition , the @xmath63-th term in the sum equals the number of outgoing @xmath58 prongs that the @xmath62 string would generate via the hanany - witten effect were it moved through the brane , so that it no longer crosses the branch cut .
as shown in fig .
[ stringcrosscut ] , upon crossing the branch cut @xmath54 in the counterclockwise direction , the @xmath62 string turns into the @xmath66 string where @xcite @xmath67 and indeed @xmath68 this confirms that the @xmath62 piece of junction can be moved across the @xmath52 brane to a position where it does not intersect @xmath54 , by adding @xmath69 _ outgoing _ prongs at @xmath52 that reach the newly located piece of junction .
similarly , a string in the junction that does not intersect @xmath54 could be moved across the brane , creating in the process a number of prongs . in
can be readily verified that in this case eqn .
( [ invch ] ) would give two canceling contributions .
note in particular that a crossing transformation may give rise to more than one prong .
equation ( [ invch ] ) shows that the contributions from the cut @xmath54 only depend on the total @xmath1 charge being transported across it .
as we deform @xmath54 continuously , the @xmath1 charge transported across it will not change .
this is clearly true even as we cross junction points since charge is conserved at these points .
we conclude that the @xmath17 charge defined above is invariant under deformations of the cut @xmath54 , as long as the deformation does not require that the @xmath54 cut cross another 7-brane .
the above discussion of invariant charges suggests how to define a canonical presentation for any junction .
the idea is to transform any given junction into an equivalent junction that does not intersect any branch cut .
this is clearly possible since any specific intersection with a branch cut can be transformed away at the cost of introducing some prongs , and this can be done with every intersection . before doing this in detail
we emphasize that we also present the branes in a canonical fashion . following the conventions of @xcite we use three types of branes , a - branes , b - branes and c - branes . in our pictures , from left to right a - branes appear first , then b - branes and finally c - branes .
all branes have their cuts going downwards vertically .
a - branes are represented by heavy dots , b - branes by empty circles , and c - branes by empty squares ( see fig . [ canonpres ] ) .
our conventions for monodromies are those of @xcite : @xmath70 : \,\,k_a = & m_{1,0}^{-1 } = t^{-1 } = \pmatrix{1 & -1 \cr 0 & 1 } \
, , \nonumber\\ { \bf b}= & [ 1,-1 ] : k_b = & m_{1,-1}^{-1 } = s t^{2 } = \pmatrix{0 & -1 \cr 1 & 2 } \ , , \\ { \bf c}= & [ 1,1 ] : \,\ , k_c = & m_{1,1}^{-1 } = t^{2}s = \pmatrix{2 & -1 \cr 1 & 0 } \,,\nonumber\end{aligned}\ ] ] where @xmath71 is the matrix @xmath72 we can now go back to the canonical presentation of junctions .
consider , for example , the junction shown in fig .
[ canonpres](a ) .
after transformation the junction will look as in fig .
[ canonpres](b ) . at every brane
there is a set of prongs that converge towards a region where there could be a complicated graph .
finally , there can be an outgoing string carrying he total charge of the configuration .
now we can allow the length of all the strings in the complicated region to approach zero , reducing the graph to a single junction point where all the prongs emerging from the various branes meet , as shown in fig .
[ canonpres](c ) .
this configuration could be taken as the canonical presentation but , for convenience , we will do a further step by introducing sub - junction points .
all prongs of the leftmost set of mutually local branes will join together forming a single string which is then joined by the string resulting from the prongs of the second set of mutually local branes , and so on . this is shown in fig . [ canonpres](d ) .
we will call this the canonical presentation . by construction ,
any junction is equivalent to its canonical presentation . as emphasized before
, we do not address the issue of whether any the junction has a bps representative .
furthermore , if it has one , it need not be the junction in the canonical presentation .
@xmath73 note that the canonical presentation is determined _ uniquely _ by the invariant charges that characterize the junction .
in fact , the invariant charge for any brane is precisely the number of outgoing prongs in the canonical presentation .
it follows that any two junctions with the same invariant charges are equivalent because each is equivalent to the same canonical presentation .
moreover , since the charges can not be changed by the allowed transformations , two junctions with different charges are necessarily inequivalent .
all in all , the invariant charges give a complete classification of all possible equivalence classes of junctions .
the canonical presentation suggests a natural way of building junctions using a basis of _ strings_.
let us label the branes using an index @xmath74 , where @xmath3 is the total number of branes in the configuration .
for a brane with label @xmath52 and of type @xmath53 $ ] , we define the string @xmath18 as the outgoing @xmath75 string starting at the brane and going off to infinity without crossing any branch cut .
this is a very special and simple junction , with invariant charges @xmath76 the string @xmath77 is the same string with reversed orientation .
let us define what is meant by adding junctions .
given two junctions @xmath78 and @xmath79 , we say that @xmath80 is the junction with charges @xmath81 for all branes @xmath52 in the set of branes .
in other words , we simply add the prongs in the canonical presentation .
this operation is illustrated for two special examples in fig .
[ twojunctions ] . in the first example
we show the addition @xmath82 of two mutually nonlocal strings .
note that in the canonical presentation this is a three string junction with outgoing prongs at the @xmath45 and @xmath46 branes and one outgoing string carrying the total charge @xmath83 . in the second example we consider @xmath84 where @xmath45 and @xmath85 denote two branes of the same type .
this time when we form the canonical presentation for the sum the total outgoing charge vanishes .
this junction is simply a string departing from the @xmath45 brane and ending on the @xmath85 brane .
@xmath86 it follows from these considerations that any junction @xmath20 can be written as a sum of basis strings , where the expansion coefficients are the invariant charges .
we have @xmath87 which is clearly in agreement with equation ( [ addjunction ] ) by virtue of eq .
( [ basisst ] ) . since the charges @xmath17 must be integers
, we have a _ lattice of junctions _
, with the set of basis strings serving as basis vectors .
each point in this lattice represents an equivalence class of junctions .
since each seven - brane of the background provides a basis string , the dimensionality of the junction lattice equals the number of seven - branes . as we proceed in our exploration of the lattice of junctions
, we will need objects having non - integral expansion coefficients in the basis of strings .
these we shall call _ improper junctions _ , to distinguish them from the proper junctions which can correspond to physical states .
in this section we consider some additional structure on the lattice of junctions .
we will introduce a symmetric bilinear form @xmath88 , _
i.e. _ an inner product , or a metric
. we will define this inner product by giving all inner products of basis strings .
this inner product will be sometimes referred to as an intersection bilinear , since it originates from the intersection bilinear on the two - dimensional submanifolds of @xmath89 which are the f - theory ( or m - theory ) lifts of the junctions .
the discussion presented here focuses on the junctions and not on their lifts .
we will call @xmath90 the self - intersection invariant of the junction @xmath20 .
the relevance of the metric in the junction lattice is that it is related to the metric in weight lattice of the associated lie algebra .
some of this relation will be explored in this section .
a related discussion of self - intersection with a different application in mind was given in @xcite .
after a brief review of the relevant notions in lie algebras , we show how to identify the roots of the algebra for the various brane configurations . using the intersection bilinear
we can readily show that the inequivalent junctions of self - intersection @xmath25 are equal in number to the roots of the expected lie algebra .
the junctions ( or strings ) corresponding to simple roots are easily identified in each case . since all roots are represented by junctions of self - intersection @xmath25 , which is also the self - intersection number of any jordan open string joining two different branes , we are led to hypothesize that all roots of the @xmath91 series of algebras have presentations as jordan open strings .
many of these were known previously @xcite , except for a class of roots in @xmath92 and another class in @xmath31 .
we construct explicitly jordan open strings representing these roots , thus confirming the hypothesis .
the intersection bilinear , defined on the equivalence classes of junctions , will be deduced by starting with an initial assignment for the self - intersection of basis strings , and then exploring the implications of bilinearity and invariance under junction transformations .
we declare the self - intersection of any basis string @xmath93 to be minus one , @xmath94 to find out the value we ought to assign to the self - intersection of a three string junction , we can consider ( as in the previous section ) an @xmath95 string crossing the cut of a @xmath6 $ ] brane . in this configuration
the string clearly has no self - intersection .
if we move the string through the brane , it no longer crosses the cut but we obtain prongs beginning on the brane and ending on the string , forming some equivalent junction . since we expect the self - intersection to be invariant under junction transformations , the various contributions must cancel .
the contribution from the prongs is @xmath96 $ ] , where @xmath97 is the number of prongs emerging from the brane ( see eq .
( [ confirmhw ] ) ) . to cancel this
we must have a contribution to the self - intersection arising from the junction point as follows .
consider the three string junction @xmath98 defined as the joining of @xmath99 strings , with @xmath100 , where all strings are ingoing ( or outgoing ) and the label @xmath101 increases ( cyclically ) as we go around the junction in the counterclockwise direction .
then the contribution to the self - intersection from the junction point is given as @xmath102 this result is independent of @xmath101 ( with @xmath103 and @xmath104 ) and can be readily verified to be invariant under @xmath105 transformations of all the strings making up the junction .
the self - intersection of a larger junction is the sum of the contributions from the various three - string junctions constituting the large junction . to this
must be added the contribution of the various prongs of the junction ending on branes .
one readily verifies that this prescription indeed gives a total self - intersection of zero for the situation described in the previous paragraph .
this result is actually sufficient to deduce the intersection matrix in the basis of strings .
we consider the three standard types of branes , @xmath106 and @xmath47 , and corresponding basis strings @xmath107 and @xmath108 .
consider now the junction @xmath109 , representing a basis string departing the @xmath110 brane that joins at a junction point a basis string departing the @xmath111 string , with @xmath112 .
its self - intersection has a contribution of @xmath25 from the two prongs at the two branes , and zero contribution from the junction point since the two strings joining there are of the same charge and the determinant in ( [ sijunc ] ) vanishes .
thus @xmath113 showing that @xmath114 when @xmath115 .
in general we conclude that @xmath116 consider now the junction @xmath117 made by combining an arbitrary @xmath118 string and an arbitrary @xmath119 string . it is important here to notice that using the canonical presentation , at the junction point the counterclockwise cyclic ordering of incoming strings is @xmath120 .
thus the self - intersection is @xmath121 on the other hand @xmath122 and we therefore deduce that @xmath123 .
exactly analogous considerations give us @xmath124 these relations define completely the intersection matrix for basis strings . if @xmath125 and @xmath126 denote the number of @xmath106 and @xmath47 branes respectively , we can define the set of basis strings @xmath127 with @xmath128 , as @xmath129 with this notation , the intersection metric is given as a matrix @xmath130 for any pair of junctions @xmath20 and @xmath131 we have @xmath132 for the case of @xmath14 , where @xmath133 , the metric is readily written with the help of ( [ ip ] ) and ( [ matrixsi ] ) @xmath134 as we can see , the last two rows just differ by a sign and thus this metric is degenerate .
this will be the case as long as @xmath135 , and therefore the metric will be degenerate for brane configurations leading to @xmath9 enhanced symmetry ( @xmath136 . for the above matrix , in addition , the sum of the first four rows is proportional to the fifth one , and thus the matrix has two zero eigenvalues ( and four negative eigenvalues , three of them equal to minus one and one equal to minus three ) . as we shall see , there will be two linearly independent string junctions that represent for @xmath14 the two eigenvectors with eigenvalue zero .
these junctions have zero intersection number with any string junction . for @xmath9 with @xmath24
there is only one zero eigenvalue and one null junction . for the case of @xmath137
the metric is given by @xmath138 this metric is nondegenerate ; its determinant equals @xmath139 .
it does not have definite signature , however , six eigenvalues are negative and two are positive . for the @xmath140 series ,
in general , one finds @xmath3 negative eigenvalues and two positive eigenvalues , as will be confirmed later ; the metrics for @xmath92 and @xmath31 are obvious generalizations of that for @xmath137 .
we now give the expression for the self - intersection of an arbitrary junction in terms of its invariant charges .
we write @xmath141 which is general enough for the present applications . using ( [ ip ] )
we obtain @xmath142 it is possible to simplify this expression by means of the total asymptotic charge of the junction .
using @xmath143 we find @xmath144 this expression will enable us to show that the junctions of self - intersection number @xmath25 are in one - to - one correspondence with the roots of the expected lie algebras .
@xmath145 in computing the intersection number of two junctions directly from their pictures , rather than from the general expression ( [ usethis ] ) one must consider the extra contributions that arise when strings simply cross without forming a junction .
one way to derive this contribution is to reconsider the contribution of a junction point to the self - intersection , and to interpret this contribution , as usual , as the intersection of two slightly displaced versions of the junction ( see fig . [ splitjunctions ] ) .
we can see that the contribution from the junction point is equal to the contribution due to the crossing of the @xmath146 and @xmath147 strings .
we can make this unambiguous as follows . at a crossing point
we have two ingoing and two outgoing strings .
pick either of these two groups and label the strings one and two , with string two following directly string one by counterclockwise rotation . the contribution
to intersection is @xmath148 to compute @xmath149 from the pictures of the @xmath20 and @xmath131 junctions we draw the junctions so that no junction points coincide , and we let the asymptotic strings join at a junction point ( an asymptotic junction ) to form a single asymptotic string .
the intersection number has the following contributions : ( i ) @xmath150 for each brane where @xmath20 and @xmath131 have @xmath3 and @xmath151 outgoing prongs respectively , ( ii ) a contribution given by ( [ crosscont ] ) for each crossing of strings , ( iii ) a contribution given by_one half _ of the result quoted in ( [ sijunc ] ) for the asymptotic junction .
to compute the self - intersection @xmath90 from the picture of the junction @xmath20 we add : ( i ) @xmath152 for each brane where @xmath20 has @xmath3 prongs , ( ii ) a contribution given by ( [ crosscont ] ) for each crossing of strings , ( iii ) a contribution given by ( [ sijunc ] ) for each junction point .
we will not give a detailed derivation of these rules here . in order to fix notation
, we now give a brief review of some basic facts about simple lie algebras .
for more background , see @xcite .
consider a compact simple lie algebra @xmath153 of rank @xmath154 with cartan generators @xmath155 , @xmath156 satisfying @xmath157 = 0 $ ] .
the remaining generators @xmath158 satisfy @xmath159 = \alpha^i\ , e_\alpha$ ] , and are thus characterized by the @xmath154 eigenvalues @xmath160 defining the _ root vector _ @xmath161 in an @xmath154-dimensional euclidean space @xmath162 .
the finite set @xmath163 of all roots is called the root system .
the commutator @xmath164 $ ] can be nonzero only if @xmath165 , or if @xmath166 .
the _ base _
@xmath167 is a subset of @xmath163 whose elements are @xmath154 _ simple roots _ forming a basis for @xmath162 , with the special property that all roots in @xmath163 can be expanded in coefficients that are either all negative or all positive .
we will be concerned henceforth only with the simply laced algebras @xmath168 and @xmath169 and @xmath31 , for which all roots have the same length squared , conventionally chosen equal to two .
a state transforming under an arbitrary representation of @xmath153 is characterized by a weight vector @xmath170 , where @xmath171 is the eigenvalue of @xmath172 on the state .
each representation is characterized by a highest weight vector @xmath173 ; all other weights are obtained from @xmath173 by subtracting simple roots . for a generic weight @xmath174 and a given simple root @xmath175 , there will exist numbers @xmath4 and @xmath5 such that @xmath176 and @xmath177 are also weights , but @xmath178 and @xmath179 are not .
then one has @xmath180 .
the cartan matrix for the algebras under consideration is a symmetric matrix with integer entries defined by @xmath181 whose information can be encoded in the dynkin diagram , which for @xmath0 algebras is quite simple ; nodes correspond to simple roots ; nodes are joined by a line if the corresponding roots are at an angle of @xmath182 , and are not joined if the roots are orthogonal .
fundamental weights @xmath183 with @xmath184 are defined as dual basis vectors @xmath185 and form the _
dynkin basis_. the expansion of a weight vector in this basis reads @xmath186 with integers @xmath187 called the _
dynkin labels_. representations are characterized by the dynkin labels of the highest weight .
such labels are necessarily non - negative .
note that , by construction @xmath188 and the cartan matrix is seen to have the dynkin labels of the simple roots as its rows .
jordan open strings beginning on one brane and ending on another one of the same type manifestly have self - intersection @xmath25 and vanishing asymptotic charges .
these strings represent the familiar roots of @xmath22 algebras .
we will see that all roots , even those in enhanced symmetry algebras , are represented by junctions of self - intersection @xmath25 and vanishing asymptotic charges . in this subsection
we will search for all such junctions for the cases of @xmath22 and @xmath8 enhanced symmetry .
the simple roots are identified as a subset of the junctions whose intersection matrix is minus the cartan matrix of the corresponding lie algebra .
consider first the case of @xmath37 , which is built from @xmath3 @xmath45 branes , with no @xmath46 branes nor @xmath47 branes . in this case the self - intersection formula ( [ siformula ] ) reduces to @xmath190 in this case the vanishing asymptotic charge condition gives @xmath191 the last two equations imply that the junctions must have two a prongs one ingoing and one outgoing .
thus the complete list of solutions for the inequivalent junctions is @xmath192 making a total of @xmath193 junctions . as expected , this is precisely the number of roots of @xmath37 .
note that they all have a simple interpretation as an open string extending between two different branes .
we can identify the simple roots as @xmath194 this is in accord with the cartan matrix of @xmath37 since @xmath195 .
the dynkin diagram and the brane configuration with simple roots are shown in fig . [ dynkinsun ] . in this case
it is straightforward to see that all roots can be written as linear combinations of simple roots with coefficients that are all positive or all negative .
the highest root is @xmath196 @xmath197 consider now the case of @xmath9 which contains @xmath3 branes of type @xmath45 , one brane of type @xmath46 and one brane of type @xmath47 . in this case
the conditions for roots are @xmath198 again , we must have two prongs on two different @xmath45 branes . if the two prongs have opposite signs we must have @xmath199 . these are @xmath200 if the two prongs have the same sign , we need two more prongs of opposite signs , one going to @xmath46 and one going to @xmath47 @xmath201 the last two expressions give a total of @xmath202 roots , the expected number for @xmath9 .
it is easy to select a set of @xmath3 simple roots .
we take @xmath203 it is clear that the intersection matrix of these roots generates the dynkin diagram of @xmath9 . in particular note
that @xmath204 is the root at the vertex ; @xmath205 , and @xmath206 , as required .
the dynkin diagram and the brane configuration with simple roots are shown in fig . [ dynkinso2n ] .
the simple roots generate all roots , but this time it takes a little effort to see that the expansion coefficients are all positive or all negative . in the above
lists the roots with @xmath207 sign in front are positive roots , and the roots with @xmath208 sign in front are negative roots .
the highest root in the adjoint representation is @xmath209 @xmath210 this time we must consider the equations @xmath211 where the number @xmath3 of @xmath45 branes takes values @xmath212 for the cases of @xmath169 and @xmath31 respectively . note that a major simplification occurring in the previous subsection does not take place here : the first equation does not imply immediately a bound on the number of @xmath45 prongs allowed .
we will now do a preliminary analysis to show that indeed the above constraints do bound the number of @xmath45 prongs .
note first that ( [ siexc2 ] ) implies that the number of @xmath45 prongs must be even .
take @xmath213 and consider the case when @xmath214 .
it then follows from ( [ siexc1 ] ) that @xmath215 and therefore @xmath216 and @xmath217 have the same sign . from the ( [ siexc2 ] ) we find @xmath218 since @xmath216 and @xmath217 have the same sign , we then have @xmath219 on the other hand @xmath220 is bounded from below .
we can go a long way with a simple bound .
we can take @xmath221 , for all @xmath101 , to find @xmath222 .
on account of ( [ siexc1 ] ) this gives @xmath223 the last two inequalities can only be solved if @xmath224 for @xmath225 , the largest @xmath63 is @xmath226 .
this implies that for @xmath137 we will at most have four @xmath45 prongs .
for @xmath227 we see that @xmath228 are possible .
thus for @xmath92 we will have to consider the possibility of up to six @xmath45 prongs .
for @xmath229 it appears as if @xmath230 . actually @xmath231 is not allowed . indeed , with @xmath232 prongs and 7 a - branes @xmath233 ( three branes having two prongs and four branes each with a single prong ) .
then eq .
( [ ineq3 ] ) becomes @xmath234 which is not compatible with eq .
( [ ineq2 ] ) which now gives @xmath235 .
therefore the maximal @xmath63 for @xmath31 is @xmath236 corresponding to a maximum of eight prongs .
this is interesting since it requires that at least one @xmath45 brane have two or more prongs .
note that @xmath237 , which we would obtain by adding an additional @xmath45 brane to the @xmath31 configuration , seems to allow all numbers of @xmath45 prongs ( see ( [ ineq4 ] ) ) .
certainly the number of @xmath45 prongs is not bounded , one can readily verify that a total of @xmath238 prongs , @xmath63 on each @xmath45 brane , is compatible with @xmath239 and @xmath240 .
this phenomenon is likely to be related to the observations of imamura @xcite .
we now consider each exceptional algebra in turn . for @xmath137 ,
the relevant equations are ( [ siexc1 ] ) , ( [ siexc2 ] ) , and ( [ siexc3 ] ) with @xmath225 . in view of our previous analysis
we must consider the case of zero , two or four @xmath45 prongs .
begin with the case of no @xmath45 prongs .
we then need @xmath242 , and we get two junctions @xmath243 the case of two @xmath45 prongs is rather similar to the earlier one .
note that the two prongs can not be on the same @xmath45 brane for then @xmath244 , and the second equation can not be satisfied .
the list of satisfactory junctions with two @xmath45 prongs is @xmath245 this gives a total of @xmath246 junctions .
finally , we must consider the possibility of having four @xmath45 prongs . a short analysis of the first two constraint equations
indicate that there are no solutions unless the four prongs fall on four different @xmath45 branes .
it then follows from the first equation that @xmath247 and the only integer solutions are @xmath248 in which case @xmath249 , by use of the second equation .
it thus follows that all @xmath45 prongs must be either outgoing or ingoing .
the junctions are then seen to be of the form @xmath250 for a total of 10 junctions .
the complete solution set thus gives @xmath251 which is indeed the number of roots in @xmath137 .
we now choose a set of simple roots , @xmath252 the corresponding string junctions are shown in fig .
[ dynkine6 ] , and one can readily verify that their intersection matrix precisely coincides with the result encoded in the dynkin diagram .
it takes work to verify that all the junctions listed above arise as positive or as negative linear combinations of the simple roots , so we will not do this explicitly .
we simply note that the highest root is given by the following junction . @xmath253
all roots of @xmath137 have familiar representations where their junctions take the form of jordan open strings between two branes .
@xmath254 for @xmath92 we will now be briefer .
a new kind of junction arises from ( [ siexc1 ] ) , ( [ siexc2 ] ) , and ( [ siexc3 ] ) with @xmath227 .
again , a short computation using the first two equations shows that no solution exists unless each of the six prongs is on a different @xmath45 brane . in this case
the first equation gives @xmath256 , which together with the second equation is solved by @xmath257 or @xmath258 , together with @xmath259 for all @xmath260 .
the junctions are therefore @xmath261 it is of course straightforward to draw junctions representing the above states .
in fact , a representation of the above states as a @xmath262 open string was given in @xcite . that string , however , had self - intersections .
it is possible to find a representative that is an open string without self - intersections .
this is shown in fig .
[ jordan7 ] , where we exhibit an @xmath263 string with appropriate charges .
the simple roots of @xmath92 can be chosen as follows : @xmath264 these roots are shown in fig .
[ dynkine7 ] , and one can see that their intersection numbers generate the corresponding dynkin diagram .
the highest root of @xmath92 is @xmath265 @xmath266 @xmath267 for @xmath31 there is one interesting new possibility . as shown earlier
, it is possible to have eight @xmath45 prongs .
once more , straightforward analysis of all possibilities shows that all eight prongs must have the same sign , and that all branes have one prong , except for one that has two prongs .
therefore @xmath269 and @xmath270 .
this requires @xmath271 with @xmath259 for six out of the seven @xmath45 branes , and @xmath272 for the other one .
the roots then read @xmath273 it is possible to find open string representatives without self - intersections for the above junctions . in fig .
[ jordan8 ] we show two types of strings .
the first type , illustrated in ( a ) , represents the state indicated above with overall sign plus and @xmath274 .
a similar configuration can be used to represent the states with @xmath275 .
the state with @xmath276 requires a slightly different style of string , shown in part ( b ) of the figure .
the simple roots of @xmath31 can be chosen as follows : @xmath277 these roots are shown in fig .
[ dynkine8 ] , and one can see that their intersection numbers generate the corresponding dynkin diagram . the highest weight vector in the adjoint is given as : @xmath278 @xmath279 @xmath280
the ( proper ) junctions representing roots do not form a complete basis on the junction lattice . similarly ,
the associated dual junctions representing fundamental weights , which are not necessarily proper , do not form a complete basis . in the present section
we will introduce additional ( not necessarily proper ) junctions having the interpretation of two new fundamental weights that together with the other weights make up an extended system of weights defining a complete basis of the junction lattice algebras only one extra weight is introduced . ] .
these additional junctions carry non - vanishing asymptotic @xmath1 charges .
in fact we will see that the @xmath4 and @xmath5 labels play the role of dynkin labels for the two new junctions .
the extended system of weights has a well - defined intersection matrix , the analog of the inverse cartan matrix .
this matrix does not always have an inverse .
it does for the case of the @xmath22 and @xmath10 algebras , but it does not for the case of the @xmath8 algebras .
when the matrix is nonsingular we can introduce two new junctions , also not necessarily proper , representing roots @xmath281 and @xmath282 dual to the new weights . in this extended formalism we will be able to conjecture the correspondance between the junction and weight lattices
. moreover we will be able to relate the self - intersection number of a junction to the length of the corresponding weight vector . in the previous section we identified junctions @xmath26 for roots .
these junctions carry zero asymptotic charges . mimicking the definition of fundamental weights
, we can use the inverse cartan matrix to define `` junctions '' @xmath27 for fundamental weights as some linear combinations of the root junctions .
these will satisfy @xmath283 along with @xmath284 note that the @xmath27 are not proper junctions since ( except for @xmath31 ) these are not integer linear combinations of basis strings ; their expansion coefficients in the string basis are rational , however .
the @xmath27 , being linear combinations of the @xmath26 , carry zero asymptotic charges and do not span the entire junction lattice . we are now going to introduce one special junction @xmath285 for each @xmath22 algebra , and two special junctions @xmath286 for each algebra in the @xmath287 and @xmath43 series , to span the whole space .
these will also generically be improper .
we want these junctions to be orthogonal to the root junctions , @xmath288 this orthogonality condition suggests that these two junctions have the interpretation of two new fundamental weights .
given that the weights @xmath27 are linear combinations of the roots @xmath26 , we also have @xmath289 the @xmath27 together with @xmath286 span the junction lattice . in order to fix the basis in the @xmath286 subspace
we add one more constraint : we demand that @xmath285 and @xmath290 carry asymptotic charges @xmath291 and @xmath292 respectively , thus @xmath293 as we will see , these conditions determine uniquely the junctions @xmath285 and @xmath290 .
since the inner product of fundamental weights defines the inverse cartan matrix , we define analogously @xmath294 having this matrix will be useful to relate the multiplicities of junctions for different values of the asymptotic charges . given a junction @xmath20 it can always be expanded using the basis @xmath295 . since the weights @xmath27 carry no asymptotic charge , such expansion will read @xmath296 where @xmath35 are a set of numbers , @xmath297 , and @xmath298 .
we will show in sections 6 and 7 that constants @xmath35 are precisely the dynkin labels of the lie algebra weight vector @xmath299 associated to the junction @xmath20 ; in other words we claim that the weight vector associated to ( [ jexp ] ) is @xmath300 included in this relation is our assertion that the @xmath27 are the junctions corresponding to the fundamental weights , @xmath301 .
equations ( [ jexp ] ) and ( [ wvass ] ) fix completely the map taking the junction lattice to the weight lattice .
it now follows that the intersections of @xmath20 with @xmath285 and @xmath290 only depend on the @xmath4 and @xmath5 charges of the junction .
we find @xmath302 furthermore , we can easily calculate the intersection number of two junctions .
the self - intersection number of any junction follows from eqns .
( [ jexp ] ) , ( [ bnb ] ) and ( [ necc ] ) : @xmath303 and on account of ( [ wvass ] ) we get @xmath304 relating the self - intersection of a junction to the length of the ( claimed ) associated weight vector .
having an extended system of weights , it is natural to define an extended inverse cartan matrix @xmath305 .
construct the basis @xmath306 where the @xmath28 are @xmath307 , or just @xmath285 for the case of @xmath22 . then define @xmath308 note that in view of the orthogonality relations ( [ opnjunn ] ) this extended matrix takes the block diagonal form @xmath309 we will see that for the @xmath8 algebras the matrix @xmath310 is degenerate and we can not define its inverse . on the other hand , for the @xmath10 algebras @xmath310 is nondegenerate and we can calculate its inverse .
we can then introduce roots @xmath311 dual to the weights @xmath312 such that @xmath313 since the @xmath29 are linear combinations of the @xmath28 we have that @xmath314 we have therefore obtained an extended set of roots @xmath315 dual to the extended weights @xmath316 it now follows from ( [ jexp ] ) and the last few equations that @xmath317 these two equations can be contrasted with the relation @xmath318 following from ( [ jexp ] ) and ( [ edual ] ) . we are therefore led to define extended `` dynkin '' labels @xmath319 which enables us to rewrite ( [ jexp ] ) as @xmath320 note that this equation is simply a statement about junctions and their expansion in terms of fundamental weight junctions .
we still have to show that the coefficients @xmath35 are indeed the dynkin labels of the weight vector associated to the junction @xmath20 . in this case
the only relevant asymptotic charge is @xmath4 and therefore we introduce @xmath285 , which is given by @xmath321 and satisfies both ( [ njunn ] ) and ( [ cnjunn ] ) . for this weight
we find @xmath322 there is an associated root @xmath323 satisfying all the requisite properties .
the extended inverse cartan matrix and extended cartan matrix read @xmath324 we can use this result to compute the intersection number of a junction . using ( [ meq ] )
we find @xmath325 note that the right hand side is manifestly positive definite .
we now turn to the case of @xmath14 , which we examine explicitly before looking at the general case of @xmath9 , as its triality properties make it rather special . using the defining equations ( [ njunn ] ) and ( [ cnjunn ] )
, we find that the fundamental weights read @xmath326 we have introduced the above notation because we wish to emphasize that @xmath327 and @xmath327 are proper junctions , while @xmath285 and @xmath290 are not .
a simple computation shows that @xmath310 vanishes identically .
the junctions @xmath327 and @xmath327 are actually null junctions , where by `` null '' we mean that their inner product with any junction vanishes .
it is simple to verify that these are the unique null junctions by looking for the most general junction with vanishing intersection with all basis strings .
it is interesting to note that one can have a presentation of @xmath328 and @xmath329 with manifest zero self - intersection .
these are simply @xmath330 and @xmath331 strings that go around all branes , cutting all branch cuts but having no prongs .
the effect of the branch cuts is a monodromy @xmath332 that is responsible for giving the total outgoing charges of @xmath333 and @xmath334 respectively .
this is shown in fig .
[ nulljunctions ] .
@xmath335 given that the extended inverse cartan matrix is degenerate we are not able to introduce dual roots @xmath281 and @xmath282 .
this is not a difficulty , it just means that the @xmath1 charges of a junction can not be obtained from a computation using the intersection bilinear .
this also means that the @xmath285 and @xmath290 junctions do not contribute to the self - intersection , and we have that @xmath336 in @xmath14 the self - intersection number of a junction equals the length squared of the associated weight vector , regardless of the asymptotic charge . for the case of @xmath9
we find extended fundamental weights @xmath337 where @xmath285 is null , but @xmath290 is not . in fact
@xmath285 is the unique null junction ( up to normalization ) .
we can only introduce one new root , a root @xmath282 satisfying @xmath338 , and taking the form @xmath339 and unique up to an irrelevant additive factor proportional to @xmath340 . note that as expected , this root fails to exist for @xmath341 .
it now follows by a simple computation that @xmath342 this completes our discussion of the extended formalism for @xmath9 . for the @xmath169 and
@xmath31 algebras the extended inverse cartan matrix is nondegenerate and one can introduce dual roots .
the results for these algebras are listed below . in this case
we find @xmath343 the @xmath344 block of the cartan matrix is readily computed @xmath345 we also find roots @xmath346 finally , the self - intersection of a junction is given by @xmath347 this time we find weights @xmath348 the corresponding @xmath344 intersection matrix is @xmath349 and the associated roots read @xmath350 finally , the self - intersection invariant can be computed as @xmath351 the extended weights are given by @xmath352 the relevant cartan matrix is @xmath353 the roots read @xmath354 and contrary to the case of @xmath137 and @xmath92 , are not proper junctions . finally , the self - intersection invariant can be computed as @xmath355
having defined the extended formalism , we now turn to a search for junctions with self - intersection @xmath30 . a single string that begins on a brane and goes off
to infinity without self - crossings will have self - intersection @xmath30 , and if the brane it ends on is one of several mutually local branes , it will transform in the fundamental of a manifest subalgebra .
we would like to use self - intersection @xmath30 as a criterion for finding all the junctions that could be expected to belong to the various fundamental representations of the enhanced lie algebras .
we expect that all of these junctions will have presentations as simple jordan strings beginning on some brane and ending at infinity . naturally , we no longer constrain the asymptotic charges to vanish . here
the situation is rather trivial .
self - intersection is just @xmath356 and so @xmath357 can only be solved by junctions with a single @xmath358 , and all others vanishing .
these are just the junctions in the fundamental and antifundamental representations , and the asymptotic charge is simply @xmath359 . in principle
we could look for junctions of self - intersection @xmath30 for any asymptotic charges @xmath1 .
however , the null junctions @xmath360 found in ( [ nullj ] ) make this unnecessary . by definition
they have zero intersection with any junction ; thus for any junction @xmath20 , we can define the junction @xmath131 , @xmath361 with @xmath362 , such that @xmath363 .
although the self - intersection is unchanged under the transformation ( [ chla ] ) , the asymptotic charges shift to @xmath364 .
thus the junctions of a given self - intersection number are in one - to - one correspondence for all @xmath1 values mod 2 .
it is therefore sufficient to find the junctions for the @xmath1 values @xmath365 , @xmath291 , @xmath292 , and @xmath366 . for junctions of charges @xmath367 and self - intersection @xmath30 , we need @xmath368 , together with @xmath369 .
the first equation implies that only one @xmath370 is nonvanishing , and thus the second equation can not be solved .
there are no junctions of appropriate self - intersection for this case .
for @xmath371 the conditions read @xmath372 we can have @xmath373 for any one of the four a branes .
this gives the following eight junctions @xmath374 where we have indicated in parenthesis the representation of @xmath14 we will show these states belong to in section 6.2 .
it is possible to see explicitly that these junctions have string representatives .
we show two such representatives in fig .
[ string8v ] .
@xmath375 for @xmath376 the conditions read @xmath377 the first equation requires that there is at most one prong on any @xmath45 brane , and that all prongs must be outgoing .
the last two equations imply that the number of @xmath45 prongs can not be even .
there are solutions for one or three @xmath45 prongs , giving the following eight junctions @xmath378 once more , it is possible to show explicitly that these junctions have string representatives .
two representatives illustrating this are shown in fig .
[ string8s ] .
@xmath379 finally , for @xmath380 the conditions read @xmath381 this time the number of @xmath45 prongs can not be odd . if there are no @xmath45 prongs we get one solution , if there are two @xmath45 prongs we get six solutions , and if there are four @xmath45 prongs we get one solution .
they read @xmath382 once more we have found a total of eight junctions .
three string representatives are shown in fig .
[ string8c ] .
@xmath383 the algebra @xmath9 has a @xmath384-dimensional vector representation and two @xmath385-dimensional spinors related by an outer automorphism of the algebra ; only in the special case @xmath341 do the dimensions of all three representations agree and the duality automorphism is promoted to a triality . for @xmath386 , there is just one null junction , given in ( [ theonex ] ) .
this allows us to shift the value of @xmath4 arbitrarily without changing a junction s self - intersection .
one can shift the junction by an integer multiple of @xmath290 , @xmath387 and ask if self - intersection is preserved for specific values of @xmath388 .
we find that this requires @xmath389 , which just flips the sign of the @xmath5 charge .
this is to be expected , as we can always obtain a new junction with the same self - intersection by @xmath390 , which sends @xmath391 ; @xmath4 can then be restored to its prior value by shifting by @xmath392 .
it thus follows that to find all junctions of self - intersection @xmath30 it suffices to analyze the cases @xmath393 and @xmath394 for all nonnegative values of @xmath5 . here
we will only investigate the cases @xmath395 .
the analysis proceeds exactly as for the special case of @xmath14 , eqns .
( [ sp1 ] ) ( [ lone ] ) , with the only difference being that there are now @xmath3 a - branes .
again , it is impossible to have junctions of self - intersection @xmath30 and charges @xmath365 for all @xmath3 . for junctions of charge @xmath291
there are @xmath384 states that fit into the vector : @xmath396 for junctions with charges @xmath292 and @xmath366 we again find that there can be no more than one prong on each a - brane , and each prong must be outgoing ; the number of a prongs must be odd and even , respectively .
as @xmath3 grows , junctions with more and more prongs can be realized , resulting in representations whose dimensions go exponentially with @xmath3 , exactly what is required for the spinors : @xmath397 making up the @xmath398 and @xmath399 representations .
these junctions generalize those in eqns .
( [ strs2 ] ) , ( [ lone ] ) , having the form @xmath400 with @xmath401 and for all @xmath388 even or odd with @xmath402 .
we now find the junctions of self - intersection @xmath30 for the exceptional algebras .
we will confirm that these junctions appear in the right numbers to allow us later on to identify them with fundamental representations ( the @xmath403 and @xmath404 of @xmath137 , the @xmath405 of @xmath92 , and with a predictable subtlety , the adjoint of @xmath31 ) .
since the metrics on the corresponding junction lattices are non - degenerate , there are no null junctions to consider . hence ,
in general , we can not shift asymptotic charges arbitrarily while preserving self - intersection .
consider , anyhow , shifting a junction as @xmath406 where @xmath154 and @xmath407 must be chosen specially so that the resulting junction is proper .
calculating the self - intersection of the new junction with the help of ( [ asymeqq ] ) we find @xmath408 we will be particularly interested in the case when junctions related by ( [ shiftjpq ] ) have the same self - intersection number . given a junction @xmath20 of fixed @xmath1 charges , this will only happen , if at all , for some specific values of @xmath154 and @xmath407 .
it is clear that one can rewrite ( [ shiftjpq ] ) to view @xmath20 as the result of a transformation on @xmath131 , with integers @xmath409 and @xmath410 . as a result ,
if one has determined exhaustively the junctions @xmath411 of a given self - intersection @xmath412 with some charges @xmath1 , and one finds a transformation mapping these to junctions @xmath413 with self - intersection @xmath414 and charges @xmath415 , one can be sure that there are no other junctions with self - intersection @xmath414 and charges @xmath415 ; if there were , one could turn the transformation around and find new junctions with charges @xmath1 and self - intersection @xmath412 , contradicting the fact that these were all known .
we will not explicitly examine all possible asymptotic charges .
we consider several cases when @xmath416 and @xmath417 , and then afterwards relate these to other asymptotic charges using transformations as above .
for these asymptotic charges
the equations are @xmath418 the first equation is the self - intersection , and the other two are the constraints on the charges . here
@xmath212 for @xmath137 , @xmath92 , @xmath31 respectively .
if @xmath367 , we obtain @xmath419 , which requires @xmath420 odd , as well as @xmath421 , requiring @xmath422 to be even .
one can readily check that satisfying both conditions is impossible .
thus there are no junctions of self - intersection @xmath30 with charges @xmath365 .
now examine @xmath371 .
the equations are @xmath423 with @xmath424 .
both @xmath420 and @xmath422 must be odd , which is possible . define the total number of prongs @xmath425 .
then @xmath426 or , using ( [ eq2 ] ) @xmath427 then we square and use the triangle inequality to obtain @xmath428 now the lower bound on @xmath429 is obtained by setting @xmath430 for all @xmath101 . then @xmath431 the only way to satisfy both inequalities is @xmath432 for the three exceptional algebras we thus have @xmath433 we still must satisfy the self - intersection equation , however , and not every number of prongs permitted by the inequalities can be realized on the brane configuration . in table 1 we list the possible patterns of junctions , with the number of distinct junctions for each algebra listed at right . at the bottom we find the total number of junctions with self - intersection @xmath30 for each .
@xmath434 & conditions & @xmath137 & @xmath92 & @xmath31 + @xmath435 & @xmath436 & 5 & 6 & 7 + @xmath437 & @xmath436 , @xmath438 & 10 & 12 & 14 + @xmath439 & @xmath440 & 10 & 20 & 35 + @xmath441 & .junctions of self - intersection @xmath30 and charges @xmath291 for the exceptional algebras .
[ cols="^ " , ] & @xmath442 & & & 7 + total junctions & & 27 & 56 & 240 + [ entable ] for @xmath137 the number of junctions can precisely fill out a @xmath443 . in the case of @xmath92 , we have a number of junctions that can fill the @xmath405 .
note that although the bounds above suggested that nine a - prongs was possible for the case of @xmath92 , there is no realization on just six a - branes .
finally , for @xmath31 , we have a total of 240 junctions , and configurations with 17 a - prongs can not be realized .
the smallest representation of @xmath31 is the adjoint , the @xmath444 .
we are missing eight junctions to complete this representation .
we will see in section [ ss : dynkine8 ] that these eight junctions correspond to the cartan generators .
they can not be realized as junctions of self - intersection @xmath30 ; in fact we will see that for the present value of the asymptotic charge they are realized as a unique junction ( appearing with multiplicity eight ) of self - intersection @xmath445 .
this is perhaps not too surprising when we recall that while junctions representing roots have self - intersection @xmath25 , cartan generators are junctions of zero self - intersection .
we have considered only asymptotic charges @xmath365 and @xmath291 . as we will see in more detail later
, junctions with a given asymptotic charge can only produce representations of a certain conjugacy class .
@xmath31 has just one conjugacy class and @xmath92 has two , but for @xmath137 we have three and we would like to find a representation in the third conjugacy class , the class containing the @xmath446 .
we obtain junctions of self - intersection @xmath30 for this representation by just changing the sign of every invariant charge in the junctions for the @xmath447 ; in this case the asymptotic charge is @xmath448 .
we can obtain junctions of other asymptotic charges from the transformations ( [ shiftjpq ] ) .
for @xmath137 with initial charges @xmath291 , we find @xmath449 recall that ( [ w61 ] ) requires @xmath450 ( mod 3 ) to obtain proper junctions .
one solution is @xmath451 which produces 27 junctions with @xmath30 self - intersection and @xmath366 asymptotic charge . as we explained before ,
these exhaust the junctions at the given self - intersection and asymptotic charge . for @xmath92 with the same initial charges ,
we arrive at @xmath452 ( [ w71 ] ) , ( [ w72 ] ) require that @xmath453 ( mod 2 ) for proper junctions .
a possibility is @xmath454 , resulting in 56 junctions of asymptotic charges @xmath455 .
finally we consider @xmath31 . instead of @xmath291 ,
begin with @xmath365 .
we find @xmath456 one possibility is @xmath457 , which changes the self - intersection by @xmath458 .
this which produces the @xmath291 adjoint with @xmath30 self - intersection ( except for the cartan generators ) from the @xmath365 roots with non - cartan self - intersection @xmath25 .
a second transformation is @xmath459 , which also changes @xmath90 by 1 , resulting in another 240 junctions of self - intersection @xmath30 at charge @xmath455 .
we could also have obtained these directly from @xmath291 with @xmath460 , which does not change the self - intersection , as required .
these have been examples to illustrate the idea ; many more sets of junctions with self - intersection @xmath30 should exist .
for an arbitrary string junction associated to a collection of branes with some gauge symmetry @xmath153 , we would like to determine the corresponding weight vector @xmath174 giving the charge assignments of the string state .
hence we seek to determine the dynkin labels associated to the junction in terms of the invariant charges : @xmath461 where @xmath462 , and @xmath52 indexes the various branes .
the purpose of the present section is to determine the matrix @xmath463 and we will follow an explicit approach in doing so for the case of @xmath37 and @xmath14
. we will exhibit junctions for the fundamental representations and discuss the interplay between conjugacy classes of representations and asymptotic charges .
the strings representing the @xmath464 and @xmath16 representations of @xmath14 were derived independently in @xcite . in the next section
we will actually show that the results of the present section can be reproduced from the intersection bilinear .
we will see that @xmath465 .
thus the lie algebra properties of a string junction are encoded in its intersections with the simple roots .
we begin the analysis by focusing on the adjoint representation .
the root generators consist of strings that stretch between a pair of branes . since the cartan matrix gives the dynkin labels of the simple roots , we have from ( [ afromq ] ) : @xmath466 which is a constraint on @xmath463 .
note that the index @xmath52 runs over one more value than the index @xmath101 . in order to solve the constraint
effectively we do a change of basis .
we introduce new charges @xmath467 with @xmath468 , and @xmath469 where @xmath4 denotes the total asymptotic charge , and is also the @xmath50 charge in the @xmath470 .
since the roots carry @xmath471 , eq . ( [ yu1 ] ) in the new basis reads @xmath472 the matrix @xmath473 , on account of @xmath474 , is seen to be precisely equal to the cartan matrix @xmath475 .
thus @xmath476 is the identity matrix , and we therefore have @xmath477 the constants @xmath478 are determined by consideration of other representations .
a single string beginning on the configuration and going to infinity has @xmath3 choices of which brane to begin on , and transforms in the fundamental @xmath479 of @xmath37 .
these states have @xmath480 . identifying the highest weight vector @xmath481 of the fundamental with an outgoing string from the first @xmath45 brane ( no other possibility is consistent ) we find immediately that @xmath482 .
we therefore get @xmath483 the antifundamental is realized by strings ending on the branes ( @xmath484 ) .
note that the fundamental and antifundamental representations are composed precisely of the junctions with self - intersection @xmath30 .
we can now consider the reverse problem , finding the charges defining the junction given the dynkin labels of a state .
since there is one more charge than dynkin label eq .
( [ hjg ] ) must be supplemented by equation ( [ pfromqsun ] ) where we think of @xmath4 as an extra dynkin label . solving for the @xmath485 s in terms of the @xmath486 we find @xmath487
we discover that for a given weight vector , @xmath4 must have a certain value mod @xmath3 in order to assure the @xmath488 are integral : @xmath489 the right hand side of ( [ sunconj ] ) is a constant for every state in a @xmath37 representation .
in fact , its value labels which of the @xmath3 conjugacy classes of @xmath37 the representation belongs to .
thus we find an important constraint : the asymptotic charge @xmath490 of a junction is fixed ( mod @xmath3 ) by the conjugacy class of the representation it transforms under . as we shall see , the correspondence between @xmath50 factors and asymptotic charges , and the constraints relating them to conjugacy classes , is generic and will persist in the more complicated algebras realized on mutually nonlocal branes .
@xmath491 the @xmath14 algebra can be realized with four a branes , a b brane and a c brane .
the dynkin diagram and the brane configuration are shown in fig . [ dynkinso8 ] .
the algebra manifests itself through the regular maximal subalgebra @xmath492 where @xmath493 is the manifest semisimple symmetry of the branes and arises in the dynkin diagram by removing the root @xmath494 , whose associated weight we conventionally chose to be the highest weight of the @xmath16 representation .
we note a few useful decompositions : @xmath495 the string junctions making up the roots of @xmath14 are the twelve roots of @xmath493 given in ( [ sondirectroots ] ) , and strings which wind around the b and c branes in traveling between a - branes .
these are the roots in ( [ sonindirectroots ] ) .
the @xmath50 factor in ( [ decso8 ] ) is generated by @xmath496 where @xmath172 is the @xmath101th cartan generator of @xmath14 .
one can check that @xmath497 indeed commutes with all the generators of the @xmath493 : @xmath498 = \left ( \sum_{i=1}^4 \alpha^i \right ) e_\alpha = \vec{\alpha } \cdot \left ( \sum_{i=1}^4 \vec{e}_i \right ) e_\alpha = 2 \left ( \vec{\alpha } \cdot \vec{\omega}^4 \right ) e_\alpha = 0 \ , , \end{aligned}\ ] ] where we have used @xmath499 , and ( [ fweights ] ) .
we now relate dynkin labels to the invariant charges .
this time we note that there are two more branes than dynkin labels . at the same time we now have two nontrivial asymptotic charges @xmath4 and @xmath5 given by @xmath500 with @xmath341 . while for the @xmath2 configurations of the previous subsection @xmath4 denoted the @xmath50 generated by all the @xmath45 branes , here @xmath4 and @xmath5 are two linear combinations of three @xmath501 ; the @xmath50 generated by the all the a branes , the @xmath50 of the b brane , and the @xmath50 of the c brane .
this time we have @xmath502 with @xmath52 running over six values while @xmath101 runs over four values . just as in the @xmath37 case we change basis from the six original charges to charges @xmath503 defined by @xmath504 for the roots @xmath505 , @xmath506 , and
thus @xmath507 becomes @xmath508 in this basis .
once more , explicit consideration of the roots shows that @xmath509 and thus @xmath510 is the identity matrix .
we therefore have @xmath511 where we have included the possibility of @xmath4 and @xmath5 contributions for @xmath512 but not for the other labels .
in fact such contributions necessarily vanish .
the first three dynkin labels are those of the @xmath493 subalgebra and we already know that ( [ aso8first ] ) gives the correct values for that case . moreover
, b and c strings must be @xmath493 singlets and @xmath513 contributions to the first three dynkin labels would ruin this .
it only remains to determine the constants @xmath514 and @xmath515 .
we can do so by considering the three inequivalent @xmath516 representations , whose highest weight vectors are given by @xmath517 we have found in the previous section precisely three groups of eight junctions with self - intersection @xmath30 .
we will show that fitting them into the three available @xmath516 representations can only be done in a unique way .
consider first the states with @xmath518 , which in view of ( [ aso8first ] ) imply @xmath519 the eight states were listed in ( [ strs1 ] ) .
suppose we try to fit them to @xmath16 .
this is clearly impossible since we have a single nonvanishing @xmath370 and the first three zeroes in the the @xmath520 weights can not be obtained .
now let us try to fit them into the @xmath39 , in particular focus on the @xmath521 weights .
again since only one @xmath370 is nonvanishing , the first two zeroes imply that it must be @xmath522 . the junction @xmath523 could only represent @xmath524 requiring @xmath525 .
but then , the only junction that could possibly represent @xmath526 is @xmath527 and it does not .
the only consistent possibility is to assign the eight junctions to the @xmath38 , in which case @xmath528 must give the @xmath529 , fixing @xmath530 .
let us now try to fit another group of eight states to the @xmath16 representation .
we see that no states in ( [ strs2 ] ) can represent the weights @xmath520 .
the two states in ( [ lone ] ) , however , can .
this time , with @xmath531 we have @xmath532 . if we tried to identify @xmath533 with @xmath534
we need @xmath535 , but then the junction @xmath536 can not give @xmath537 . the other way around it works
, we identify @xmath533 with @xmath537 fixing @xmath538 .
then the junction @xmath536 gives @xmath537 .
the complete formulas for the dynkin labels of @xmath14 are therefore @xmath539 the @xmath50 charge , from eqn .
( [ so8u1 ] ) , is @xmath540 with these formulae , the @xmath541 , @xmath38 , @xmath39 , and @xmath16 can all be represented as simple strings that trace jordan curves .
the asymptotic charges associated to these strings are @xmath542 examples were shown in figs .
[ string8v ] , [ string8s ] , [ string8c ] . in fig .
[ string8v ] , the first string belongs to the @xmath543 , and the second belongs to the @xmath544 . in fig .
[ string8s ] the first string belongs to the @xmath545 and the second to the @xmath546 . in fig .
[ string8c ] we have states in the @xmath16 with asymptotic charge @xmath366 .
the first shown , a c string , is in the @xmath547 ; the second one is a representative from the @xmath548 and the third is the @xmath549 .
the reader may wonder about the symmetry of strings departing b and strings departing c. for this purpose we have included fig .
[ string8cb ] showing states in the @xmath16 representation generated by strings with asymptotic charge @xmath550 .
the first string is the b string , in the @xmath549 , the second is a representative of @xmath548 and the third is the @xmath551 .
these junctions are obtained from the @xmath366 @xmath16 via a shift by @xmath552 .
@xmath553 equation ( [ gotrep ] ) is consistent with familiar results in four dimensional @xmath13 supersymmetric @xmath554 gauge theory with four flavors @xcite .
as is well - known @xcite , this theory can be realized on the worldvolume of a d3-brane probe moving in the aaaabc background , with matter states in the worldvolume theory arising from strings which begin on 7-branes and end on the d3-brane . in agreement with ( [ gotrep ] ) ,
in addition to @xmath555 elementary hypermultiplets transforming in the @xmath556 of @xmath14 , monopole and dyon states with charges @xmath557 and @xmath558 , transforming in the @xmath39 and @xmath16 respectively were found @xcite . in general
an arbitrary @xmath559 dyon was said to transform in the eight - dimensional representation associated to @xmath560 , with @xmath561 ( mod 2 ) and @xmath562 ( mod 2 ) .
we explain this now . since the map from invariant charges to labels @xmath563 is invertible , as we explore in the next subsection , there must be two transformations of the invariant charges that change @xmath4 and @xmath5 while preserving the dynkin labels .
these are easily found by inspection .
one such transformation is @xmath564 with @xmath388 an arbitrary integer .
this changes the asymptotic charge by @xmath565 .
the other operation is @xmath566 with @xmath3 an arbitrary integer .
this changes the asymptotic charge @xmath567 .
the reader may recognize that these transformations are nothing else than the transformation ( [ chla ] ) generated by @xmath328 and @xmath329 given in ( [ nullj ] ) . in the next section we will see that these junctions do not change dynkin labels since they have zero intersection number with the roots . given a @xmath559 dyon we can therefore soak the asymptotic charge @xmath568 necessary to reduce it to one of the four canonical asymptotic charges via a transformation of the form ( [ nullj ] ) .
the conjugacy classes of @xmath14 are labeled by two integers ( mod 2 ) : @xmath569 it follows from ( [ aso8 ] ) that @xmath570 the condition that the invariant charges must be integers requires @xmath571 ( mod 2 ) , and @xmath572 ( mod 2 ) , and using ( [ so8con ] ) we have
@xmath573 this shows that the conjugacy class of an @xmath14 representation determines the required asymptotic @xmath559 charge mod 2 .
the representation can be constructed by junctions of any fixed asymptotic charge satisfying the mod 2 condition .
this is analogous to the situation for @xmath37 , where the conjugacy class determined the possible values of @xmath574 .
the above equation implies that @xmath575 are related as follows @xmath576 which can be checked to be consistent with the results of ( [ gotrep ] ) for the @xmath541 , @xmath38 , @xmath39 and @xmath16 representations . note that conjugacy classes of representations form a group under tensor product of representations .
indeed , in the present example , if we take any representation of the conjugacy class @xmath577 and we tensor it with any representation in the conjugacy class @xmath578 the result is a series of representations all of which belong to the conjugacy class @xmath579 ( mod 2 )
. thus conjugacy classes of representations combine according to the group @xmath580 .
the above result associates to each conjugacy class of representations an equivalence class of possible @xmath1 asymptotic charges .
since multiplication of representations implies addition of asymptotic charges for the corresponding junctions , consistency requires that the map from conjugacy classes to equivalence classes of @xmath1 charges define a homomorphism from the group of conjugacy classes to a group generated by the equivalence classes of asymptotic charges under addition .
note that for our particular example we actually have an isomorphism of groups .
the construction of @xmath14 generalizes readily to the case of @xmath581 , with the slight complication that the conjugacy classes are different for @xmath3 odd than for @xmath3 even .
since the analysis is so similar to that of @xmath14 , we directly give the result expressing dynkin labels in terms of invariant charges : @xmath582 the @xmath50 charge is given by @xmath583 when @xmath3 is even one may insert a factor of @xmath584 to the definition of @xmath585 to normalize the smallest value of @xmath585 to unity .
solving for the invariant charges we find @xmath586 these equations will place constraints on the asymptotic charges differently for @xmath3 even and @xmath3 odd .
this is consistent with the different conjugacy classes of @xmath9 for @xmath3 even and @xmath3 odd . for @xmath3
even , the group of conjugacy classes is @xmath580 , as given by @xmath587 which generalizes ( [ conjso8 ] ) .
one can readily check that the constraints on the asymptotic charges are still given by ( [ pqconjso8 ] ) , ( [ ccpqso8 ] ) . for @xmath3 odd , the situation is more complicated .
conjugacy is @xmath588 , given by @xmath589 unlike the case for @xmath3 even , the quantity @xmath590 is not independent , but is determined by the value of @xmath591 .
however , we still find it useful to define since it characterizes the constraint on the asymptotic charges .
we see that ( [ qaso2n ] ) requires @xmath592 however , in ( [ qbso2n ] ) and ( [ qcso2n ] ) it is the value of @xmath5 ( mod 4 ) that is relevant .
thus @xmath593 ( mod 4 ) and @xmath594 ( mod 4 ) will correspond to different conjugacy classes .
the value of @xmath4 in turn changes to compensate for the different value of @xmath5 , but since only @xmath595 appears , @xmath4 is only determined ( mod 2 ) .
the relationship is @xmath596 this corresponds to the following map between conjugacy ( given by @xmath591 ) and asymptotic charges , ( mod 2 ) for @xmath4 and ( mod 4 ) for @xmath5 : @xmath597
in previous section , we considered the @xmath22 and @xmath8 algebras and derived the formula giving the dynkin labels associated to arbitrary junctions by explicit examination of certain representations .
the formula ( [ afromq ] ) requires finding the matrix @xmath36 .
we showed that @xmath36 is fixed by requiring the the expected dynkin labels for the root junctions and for the junctions of self - intersection is@xmath30 . in this section
we will find a simple explanation for this fact using the intersection matrix .
( [ afromq ] ) will be reinterpreted @xmath598 .
we will show that proving this requires , in general , examination of at least two representations that have nontrivial asymptotic charges .
this better understanding of the relation between junction and weight lattices will allow us to complete the discussion of the exceptional algebras .
additionally , we solve for the invariant charges in terms of @xmath599 , and show that in each case , the conjugacy class of the algebra places restrictions on the asymptotic charges . to begin with ,
let us find the constraints on the matrix @xmath36 arising from the root junctions .
observe that the cartan matrix , giving the dynkin labels of the simple roots , is realized geometrically as the intersection of simple roots : @xmath600 where we also made use of ( [ afromq ] ) for the case of the adjoint .
if we expand @xmath505 in the basis of strings we find @xmath601 and from the last two equations we deduce that @xmath602 to find the general solution of the above equation consider the set of vectors associated to all simple roots : @xmath603 .
let @xmath604 be a basis for a set of vectors orthogonal to all of the @xmath605 vectors , namely @xmath606 for all @xmath607 and @xmath608 .
then the general solution of ( [ findk ] ) is @xmath609 the simplest solution is therefore @xmath610 if we take this to be the correct value for the matrix @xmath36 we can now find a simple expression for the dynkin labels of junctions .
we have @xmath611 showing that the dynkin labels of the weight associated to any junction are determined by its intersection with junctions associated to the simple roots .
this is exactly the result we said we would prove in ( [ hui ] ) .
it was said then that the constants @xmath35 in ( [ jexp ] ) are indeed the dynkin labels of the associated weight vector .
we now explain why ( [ solsol ] ) is the correct solution .
assume we have a set of junctions @xmath612 all of which have the same asymptotic charges @xmath1 .
similarly , assume we have another set of junctions @xmath613 all of which have asymptotic charge @xmath415 .
assume further that both @xmath1 and @xmath415 are not identically zero , nor proportional to each other .
finally , assume ( [ solsol ] ) identifies correctly these two sets of junctions to two representations .
under these circumstances , we will show that ( [ simsol ] ) is the only possible solution .
assume that @xmath614 maps correctly the junctions @xmath615 to the full set of weight vectors @xmath616 of a specific representation @xmath617 .
note that from ( [ gensol ] ) it follows that @xmath618 we now claim that there are two vectors @xmath619 satisfying the required orthogonality condition ( [ exiu ] ) .
these are the vectors @xmath620 and @xmath621 satisfying for arbitrary junctions @xmath622 since root junctions have zero asymptotic charges the requisite conditions ( [ exiu ] ) are satisfied . now
( [ getrid ] ) can be rewritten as @xmath623 the important thing to note here is that since all junctions have the same asymptotic charges all weight vectors @xmath624 are shifted by the same vector @xmath625 ; namely @xmath626 where @xmath627 since the set of weights of any ( irreducible ) representation @xmath617 can not be invariant under translations we conclude that @xmath628 . because the fundamental weights @xmath629 form a basis ,
if @xmath628 then @xmath630 similarly , the second set of junctions would require @xmath631 since we assumed that the @xmath1 and @xmath415 values are nonzero and not proportional , this implies that the constants @xmath632 and @xmath633 vanish identically .
this concludes our proof .
the above discussion is in agreement with our experience in the previous section . for the @xmath8 algebras we had to examine two sets of junctions with non - vanishing asymptotic charges to fix the matrix @xmath36 ( for the case of @xmath22 algebras ,
as usual , one representation sufficed ) .
we now verify that our previous results for @xmath37 and @xmath14 follow directly from ( [ solsol ] ) .
consider first @xmath37 , and examine @xmath634 .
since @xmath635 , the formula gets contributions only from @xmath636 , reproducing @xmath637 . in a similar fashion
we can confirm our results for @xmath14 .
the simple roots are given in ( [ sonsimple ] ) @xmath638 for @xmath100 we find @xmath639 since the @xmath119 and @xmath533 strings have zero intersection with the first three root junctions .
we then get @xmath640 and the only nontrivial terms are @xmath641 reproducing ( [ aso8 ] ) .
the case of @xmath9 proceeds along the same lines .
looking for junctions of self - intersection @xmath30 for the exceptional algebras , we found sets of 27 , 56 and 240 junctions for @xmath137 , @xmath92 and @xmath31 respectively .
how can we confirm that ( [ solsol ] ) associates these junctions precisely to the weight vectors of the @xmath443 , the @xmath405 , and to the non - zero weights of the @xmath444 ?
we claim that it is sufficient to prove that for each case one junction maps to a weight vector of the representation in question
. this will be simple to show , given that all weights of @xmath443 , all weights of the @xmath405 , and all non - zero weights of the @xmath444 have the same length , and no weight appears more than once in the representations .
series , as well as for the @xmath446 of @xmath137 .
we thank v. kac for bringing these facts to our attention . ]
focus on one of the algebras and the set of @xmath30 self - intersection junctions that we wish to show is associated to a representation @xmath642 .
assume there is a junction @xmath20 in the set that maps to a weight vector @xmath643 in @xmath642 .
let @xmath644 be a simple root that can be added to @xmath643 to get another weight vector @xmath645 in @xmath642 .
the linearity of ( [ solsol ] ) implies that @xmath646 is mapped to @xmath647 .
we now want to show that @xmath131 and @xmath20 have the same self - intersection .
recall that assuming ( [ solsol ] ) we showed that the self - intersection of a junction @xmath20 is related to the length squared of its associated weight vector by @xmath648 where @xmath649 is a quadratic form in the asymptotic charges , as ( [ relcone6 ] ) , ( [ relcone7 ] ) , ( [ relcone8 ] ) .
since @xmath650 has zero asymptotic charge , @xmath20 and @xmath131 have the same asymptotic charges and @xmath651 thus as long as @xmath647 has the same length as @xmath643 , @xmath131 has the same self - intersection as @xmath20 .
the claim now follows easily . starting with any weight produced by a junction , we can by the process of adding or subtracting simple roots
reach every weight in the representation @xmath642 . in the case of @xmath137 and @xmath92 , for each such weight we find a junction of self - intersection @xmath30 mapping to it . since no weight appears more than once we produce a set of dim@xmath652 different junctions of self - intersection @xmath30 that give correct dynkin labels .
since , by assumption , there are precisely dim@xmath652 junctions of self - intersection @xmath30 for the given values of @xmath1 the set of junctions produced by the iterative procedure must coincide with the junctions we found in the lattice . for the case of
@xmath31 we note that the 240 nonzero weights break into two groups , the positive roots and the negative roots .
all positive roots can be obtained from a positive root by adding and subtracting simple roots , and the same holds for the negative roots .
if we have a junction mapping to a positive root , minus that junction will map to the corresponding negative root . the above argument then can be applied to the two sets separately .
this concludes our proof .
we now turn to a discussion of each exceptional algebra individually .
the simple roots of @xmath137 are given in ( [ simpe6 ] ) .
compared to @xmath14 , there is now an additional a brane and an additional c brane .
the roots like @xmath653 produce cancellations and give @xmath654 as before .
similarly , we have @xmath655 , and obtain @xmath656 .
finally , @xmath657 is analogous to @xmath657 of @xmath14 .
we find @xmath658 we invert the matrix @xmath36 to find for the invariant charges @xmath659 again the condition that @xmath660 requires that the charges of a junction be determined by the conjugacy class of the representation .
@xmath137 has @xmath661 conjugacy determined by the quantity @xmath662 and we see that we must have @xmath663 the highest weight vector of the @xmath443 has dynkin labels @xmath664 . using ( [ e6dynkin ] ) , one can check that the junction @xmath528 , present in the set of 27 junctions of self - intersection @xmath30 , maps to this weight vector .
hence this set of 27 junctions with asymptotic charge @xmath291 is associated to a @xmath443 .
additionally , we can use the transformation ( [ e6shift ] ) to produce another @xmath443 , with asymptotic charges @xmath366 . since ( [ solsol ] ) successfully identifies these two sets of junctions with @xmath1 not proportional , by our previous arguments it is the only solution for @xmath137 .
the simple roots of @xmath92 are given in eqn .
( [ simpe7 ] ) .
there is one more a - brane than in @xmath137 .
the roots have the same basic structure , and similar calculations lead us to the result @xmath665 as we did for @xmath137 , we can invert @xmath36 to find @xmath666 @xmath92 has @xmath667 conjugacy , given by @xmath668 and indeed we see for the invariant charges to be integral we require @xmath669 the highest weight of the @xmath405 has dynkin labels @xmath670 . as one can verify using ( [ e7dynkin ] ) , the junction @xmath671 with asymptotic charges @xmath291 , maps to this weight vector .
this junction is among the set of 56 with self - intersection @xmath30 , and implies that this set is identified with the @xmath405 .
( [ e7shift ] ) can be used to produce another @xmath405 with asymptotic charges @xmath455 , and these two representations together establish that ( [ solsol ] ) is the only solution for @xmath92 as well .
the simple roots of @xmath31 are given in eqn .
( [ simpe8 ] ) .
there is now a total of 7 a - branes .
almost identical calculations give @xmath672 again we can invert @xmath36 , to obtain for the invariant charges @xmath673 @xmath31 has just one conjugacy class , and accordingly , there is no condition on the asymptotic charges . as discussed before ,
there are 240 junctions with asymptotic charge @xmath291 and self - intersection @xmath30 .
one of these corresponds to the highest weight @xmath674 , @xmath675 and following our earlier arguments this is sufficient to establish that these junctions are associated to the weights of the @xmath444 with non - zero length .
since the roots of @xmath31 have length squared 2 and the cartan generators have zero length squared , by ( [ jfromlambda ] ) we expect the remaining 8 weights to be associated to a junction with self - intersection @xmath445 .
the common weight vector has vanishing dynkin labels , which requires @xmath676 , @xmath677 , and @xmath678 furthermore we have for general asymptotic charges @xmath679 there is a unique solution for each @xmath1 . for our case of @xmath291
we find @xmath680 one can readily check that the self - intersection of this junction is indeed @xmath445 . using ( [ e8shift ] )
we can produce another @xmath444 , this one with asymptotic charges @xmath455 , and can use this @xmath444 together with the previous one to establish that ( [ solsol ] ) is the only possible relation for @xmath31 .
thus we have proved that ( [ solsol ] ) is the unique solution for all simply laced lie algebras .
in this paper we have examined the lie algebra representations arising from string junctions ending on a background of branes . on the space of junctions
we have defined equivalence classes .
two junctions belong to the same class if they can be transformed into each other by brane crossings and continuous deformations .
each equivalence class has a canonical representative , where the junction does not cross the branch cut of any brane .
the space of equivalence classes has the structure of a lattice , with generators called basis strings .
a general equivalence class is constructed by integer linear combinations of basis strings .
this lattice is equipped with a metric , _
i.e. _ a bilinear symmetric inner product , arising from intersection invariants .
this metric is positive definite for the @xmath22 series , degenerate for the @xmath8 series , and nondegenerate but indefinite for the @xmath10 series .
additionally , we have elucidated the relation between the lattice of junctions on this background , and the weight lattice of the lie algebra . for the @xmath22 series
the junction lattice has one more dimension than the weight lattice , while for the @xmath8 and @xmath10 series the junction lattice has two more dimensions than the corresponding weight lattice . in these latter cases ,
a choice of asymptotic @xmath1 charges selects a junction sub - lattice of codimension two which can be identified with the weight lattice of one conjugacy class of representations of the algebra . a given weight vector in this conjugacy class is thus represented as a junction with the chosen @xmath1 .
for any specific conjugacy class , the weight sub - lattice can actually be identified with a set of @xmath1 junction sub - lattices .
the set of @xmath1 values associated to each conjugacy class defines an equivalence class ; this association being an isomorphism between the group of conjugacy classes and the group of @xmath1 equivalence classes under addition . the conjugacy class of the adjoint maps to the equivalence class containing the asymptotic charge @xmath365 .
not only are junction and weight lattices related , but their metrics are related as well .
the identification of the @xmath365 junction sub - lattice to the root lattice maps one metric into the other .
for other @xmath1 sublattices the two metrics differ by a quadratic form in @xmath4 and @xmath5 .
moreover , just as the dynkin labels of a given weight are found by inner product with the simple roots , they are also given by the inner product of the associated junction with the @xmath365 junctions representing the simple roots .
it is useful to think of the entire junction space as isomorphic to an extended weight lattice , with additional fundamental weights associated to the asymptotic charges .
the additional weights are also related to the @xmath50 generators of the configuration that are not part of the enhanced gauge algebra .
important questions remain .
a full systematic analysis of the junction lattice , giving for example the states and representations for every possible intersection number , is missing .
while in the standard weight lattice a given weight vector can appear in an arbitrary number of representations , one feels that the junction lattice , having more room in it , could organize states and representations in a less degenerate way .
note that the junction lattice is in some sense more basic than the lattice of homology classes of surfaces , the familiar object typically used to see the emergence of lie algebras ( see @xcite ) .
a basis string , when lifted to the elliptic fiber , is essentially half of a two - sphere .
we have not explored @xmath681-theory or @xmath682-theory viewpoints , nor attempted to relate our work to the geometrical engineering program @xcite . the significance of the extended root system is not completely clear . as presently formulated , the extended cartan matrix is block diagonal with zero inner product between the new and old simple roots .
there may be other relevant constructions .
our work has been largely lie - algebraic , and has not concentrated on the string - theoretic realization of a junction . in particular ,
we have not attempted to determine which representations of a given symmetry algebra have associated bps junctions . while non - bps stable junctions are also of interest @xcite ,
bps states are easily identified in the string spectrum and a full understanding of their properties should be feasible .
it is known that two different holomorphic submanifolds must have a non - negative intersection number ( see , for example @xcite ) ; this condition , used in @xcite , could be of help in the present context as well .
if a given representation is known to be realizable by bps junctions , it remains to be shown that each state is uniquely realized by one and only one representative in the corresponding equivalence class of junctions for given values of the moduli .
these ideas could be relevant to elucidate the physics of @xmath13 four dimensional theories with exceptional flavor symmetries . by giving the general relation between junctions and weight vectors we have made precise and extended in a significant way the usual picture of open strings stretched between branes as representing particular generators of a gauge algebra .
perhaps our methods will find further applications in other settings , and help find generalized brane constructions giving more exotic gauge algebras and representations .
.4 in we would like to thank to t. hauer and a. iqbal for many illuminating discussions .
we are also grateful to m. gaberdiel for a critical reading of the manuscript and many useful suggestions , and to c. vafa for helpful comments .
finally , we wish to thank m. stock for preparing the figures and for technical help with latex .
in the brane configuration of @xmath23 the manifest subalgebra ( in the brane sense ) is @xmath683 , and for the brane configuration of @xmath44 , the manifest subalgebra is @xmath48 .
our goal here is to give the dynkin labels of the manifest subalgebras in terms of the dynkin labels @xmath684 of the enhanced algebras .
this is a standard embedding problem , specified by indicating the relation of the simple roots of the subalgebras to the simple roots of full algebras . in all of our cases
the manifest subalgebra has the same rank as the enhanced algebra @xmath153 and can be obtained from the enhanced algebra through a series of maximal regular subgroups @xcite . because the rank never changes , the sought - after transformation of dynkin labels is nothing but a change of basis in weight space .
the normalization of the @xmath50 charge is conventional . for @xmath685 ,
as outlined previously , the decomposition occurs by removing the fourth simple root and replacing it with the @xmath50 given in eqn .
( [ u1so8 ] ) . letting @xmath686
denote the dynkin labels of @xmath493 and the @xmath50 charge , we find @xmath687 k. hashimoto , h. hata , n. sasakura , _
3-string junction and bps saturated solutions in su(3 ) supersymmetric yang - mills theory _ , hep - th/9803127 , k. hashimoto , h. hata , n. sasakura , _ multi - pronged strings and bps saturated solutions in su(3 ) supersymmetric yang - mills theory _ , hep - th/9803127 .
5d field theories and m theory _ , hep - th/9705031 ; o. aharony , a. hanany , b. kol , _ webs of @xmath1 5-branes , five dimensional field theories and grid diagrams _ , hep - th/9710116 ; b. kol , j. rahmfeld , _ bps spectrum of @xmath701 dimensional field theories , @xmath1 webs and curve counting _ , hep - th/9801067 . c. bachas , m. r. douglas , m. b. green , _ anomalous creation of branes _ , hep - th/9705074;u .
h. danielsson , g. ferretti , i. r. klebanov , _ creation of fundamental strings by crossing d - branes _ ,
lett . * 79 * , 1984 ( 1997 ) , hep - th/9705084 ; o. bergman , m. r. gaberdiel , g. lifschytz , _ branes , orientifolds and the creation of elementary strings _ , nucl . phys .
* b 509 * , 194 ( 1998 ) , hep - th/9705130 .
r. n. cahn , _ semi - simple lie algebras and their representations _ , ( benjamin / cummings , reading , 1984 ) ; j. e. humphreys , _ introduction to lie algebras and representation theory _ , ( springer - verlag , new york , 1972 ) ; r. slansky , _ group theory for unified model building _ , phys
, 1 ( 1981 ) . | we consider string junctions with endpoints on a set of branes of iib string theory defining an @xmath0-type gauge lie algebra .
we show how to characterize uniquely equivalence classes of junctions related by string / brane crossing through invariant charges that count the effective number of prongs ending on each brane .
each equivalence class defines a point on a lattice of junctions .
we define a metric on this lattice arising from the intersection pairing of junctions , and use self - intersection to identify junctions in the adjoint and fundamental representations of all @xmath0 algebras .
this information suffices to determine the relation between junction lattices and the lie - algebra weight lattices .
arbitrary representations are built by allowing junctions with asymptotic @xmath1 charges , on which the group of conjugacy classes of representations is represented additively .
one can view the @xmath1 asymptotic charges as dynkin labels associated to two new fundamental weight vectors .
8.5 in 6.5 in -.35 in = cmss10 = cmssq8 = cmssbx10 = cmssbx10 scaled 1200 addtoresetequationsection boxedeps.tex 5_5 l # 1 # 1 | # 1| # 1 # 1#1 # 1#2#1 | # 2 # 1#2(#1 # 2 ) _ 2zsl(2 , ) # 1#2#1#2 _ 0.8 cm oliver dewolfe and barton zwiebach 0.2 cm _ center for theoretical physics , + laboratory for nuclear science , + department of physics + massachusetts institute of technology + cambridge , massachusetts 02139 , u.s.a . _ |
the phenomenon of outflowing mass from the cores of active galactic nuclei ( agns ) has been well established by now .
typical velocities of a few hundreds to a few thousands ^-1
kms@xmath0 have been measured in numerous objects ( crenshaw et al . 2003 and references therein ) .
recent studies of the x - ray spectra of certain quasars have led to claims of much higher outflow velocities reaching a significant fraction of the speed of light . in apm08279
+ 5255 , chartas et al .
( 2002 ) claim to find absorption lines of he@xmath2 outflowing at speeds of @xmath3 and @xmath4 .
hasinger et al .
( 2002 ) using different instruments prefer a more conservative interpretation for apm08279 + 5255 , by which the x - ray wind is much slower and consistent with the well known , uv broad absorption line wind of that source , outflowing at velocities of up to 12000 ^-1 kms@xmath0 . in pg1115
+ 080 , chartas et al . ( 2003 ) find two x - ray absorption systems which they attribute to he@xmath2 and hence to outflow velocities of @xmath5 and @xmath6 .
all of these measurements , however , were carried out using spectra obtained with ccd cameras and hence at moderate spectral resolving powers of @xmath7 . using the simultaneous observing mode of _ xmm - newton _ , which operates both ccd cameras and reflection gratings ( @xmath8 up to 500 ) simultaneously , high resolution x - ray spectra for several quasars have been obtained . for pg1211 + 143 pounds
( 2003a ) find a rich , well resolved spectrum featuring absorption lines of several ions , which they interpret as due to an outflow of @xmath1 24000 ^-1 kms@xmath0 .
a similar interpretation was applied to a similar observation of pg0844 + 349 , where pounds et al .
( 2003b ) report even higher velocities reaching @xmath1 60000 ^-1 kms@xmath0 . in ngc4051 ,
pounds et al .
( 2004 ) find a single absorption line at @xmath1 7.1 kev , which they suggest may be ly@xmath2 at an outflow velocity of @xmath1 6500 ^-1 kms@xmath0 , or the he@xmath2 resonance absorption line of in which case the outflow velocity is @xmath1 16500 ^-1 kms@xmath0 . yet another ultra - high - velocity ( uhv , i.e. , sub-@xmath9 ) wind of 50000 ^-1 kms@xmath0 was reported by reeves et al .
( 2003 ) for pds 456 . in all of these sources , the inferred hydrogen column densities through
the wind is of the order of 10@xmath10 , which is about an order of magnitude higher than the typical values measured for the nearby seyfert sources . if indeed uhv outflows are common to bright quasars , this could have far reaching implications on our understanding of agn winds and agns in general .
for instance , if these winds carry a significant amount of mass as the high column densities may suggest , they would alter our estimates of the metal enrichment of the intergalactic medium by quasars .
it remains to be shown theoretically what mechanism ( e.g. , radiation pressure ) can drive these intense winds . since the amount of mass in the wind is not well constrained , it is still unclear what effect it may have on the energy budget of the agn .
king & pounds ( 2003 ) note that uhv winds have been found mostly for agns accreting near their eddington limit .
they provide a theory by which the uhv outflows are optically thick producing an effective photosphere , which is also responsible for the uv blackbody and soft x - ray ( excess ) continuum emission observed for these sources . due to the potentially profound importance of the uhv wind phenomenon , in the present work
, we re - analyze the _ xmm - newton _ observation of pg1211 + 143first studied in pounds et al .
( 2003a ) . using a significantly different method
, we propose alternative identifications of the absorption lines detected by pounds et al .
the present analysis suggests that the outflow velocity of the absorber in pg1211 + 143 may be more similar to the common outflows found in seyfert galaxies . in 2 and
3 we describe the data reduction and the data analysis , respectively , and in 4 we discuss our model and results . in
5 we give a short summary of our study . throughout this paper
we use for pg1211 + 143 a redshift of @xmath11 ( marziani et al .
1996 ) and a galactic absorption column density of @xmath12 ( murphy et al .
the bolometric luminosity of pg1211 + 143 is estimated to be about @xmath13 ^-1 erg s@xmath0 ( kaspi et al .
pg1211 + 143 was observed with _ xmm - newton _ during 2001 june 15 for about 55 ks .
we retrieved the data for this observation from the _ xmm - newton _ archive and reduced them using the science analysis system ( sas v5.3.0 ) in the standard processing chains as described in the data analysis threads and the abc guide to _ xmm - newton _ data analysis .
the epic ( ccd ) data were processed using sas in the standard way .
source data were extracted from circular regions of radius 60@xmath14 .
the epic - pn was operated in the large window mode resulting in a net exposure time of 50 ks .
the observed count rate ( @xmath13.7 countss@xmath0 before background subtraction ) was well below the pileup threshold of 12 countss@xmath0 ( see the _ xmm - newton_users handbook ) . for statistical purposes , we binned the spectra to have at least 25 counts per bin .
the rgs1 and rgs2 were operated in the standard spectroscopy mode resulting in an exposure time of @xmath15 ks .
background subtraction was performed with the sas using regions adjacent to those containing the source in the spatial and spectral domains .
the spectra were extracted into uniform bins of @xmath10.04 ( which is about the rgs resolution and is 4 times the default bin width ) in order to increase the signal - to - noise ratio ( s / n ) . for the purpose of modeling narrow absorption lines
, this rebinning method is better than the method used by pounds et al .
( 2003a ) of rebinning the spectrum to a minimum of 20 counts per bin , which distorts the spectrum especially around low - count - rate absorption troughs .
to flux calibrate the rgs spectra we divided the count spectrum of each instrument by its exposure time and its effective area at each wavelength .
each flux - calibrated spectrum was corrected for galactic absorption and the two spectra were combined into an error - weighted mean . at wavelengths where the rgs2 bins did not match exactly the wavelength of the rgs1 bins , we interpolated the rgs2 data to enable the averaging .
we carefully checked that the features seen in the combined spectrum appear in both rgs1 and rgs2 .
in several pixels which show large deviation in counts due to incomplete rejection of some warm pixels ( j. kaastra , private communication ) we used the other rgs .
the sky - subtracted combined rgs spectrum has in total @xmath16 counts and its s / n ranges from @xmath17 around 8to @xmath18 around 18 with an average of 3 .
statistics in the second order of refraction are insufficient , hence we did not include it in our analysis . overall , the present data reduction agrees well with that of pounds et al .
( 2003a ) except for a few minor features which appear to be slightly different between the two reductions , for example , around 20.4 and around 17.48 in the rest frame of the source .
we attribute these discrepancies to the different binning methods used and to the averaging of rgs1 and rgs2 in this work .
we also note an unexplained @xmath19 wavelength shift between the pounds et al .
( 2003a ) data set and our reduction , but this does not affect the analysis nor the main results .
we use xspec ( v.11.1.0 ) to fit the epic - pn data .
we first fit the ( line - free ) rest - frame 25 kev energy range with a simple power law .
the best fitted power law has a photon index of @xmath20 and a normalization of @xmath21 ph@xmath22s@xmath0kev@xmath0 and gives @xmath23 for 487 degrees - of - freedom ( d.o.f . ) . extrapolating this power law up to a rest - frame energy of 11 kev
, we find a flux excess above the power law at around 6.4 kev , which is indicative of an iron k@xmath2 line , and a flux deficit below the power law at energies above 7 kev .
we add to the model a gaussian emission line to account for the fek@xmath2 line and a photoelectric absorption edge to account for the deficit .
fitting for all parameters simultaneously , we find the best - fit gaussian line center is at @xmath24 kev ( or @xmath25 kev in the rest frame ) and a line width ( @xmath26 ) of @xmath27 kev .
the total flux in the line is @xmath28 ^-2s^-1 ph@xmath22s@xmath0 . for the edge
, we find a threshold energy of @xmath29 kev , which is translated to a rest frame energy of @xmath30 kev .
the optical depth at the edge is @xmath31 .
the power law model with the gaussian line and the absorption edge gives a @xmath32 for 613 d.o.f .
this model is plotted in figure [ pnmodel ] where we show the model , both folded through the instrument and fluxed ( i.e. , unfolded ) .
we stress that this edge does not necessarily contradict the presence of the line detected by pounds et al .
( 2003a ) , since k@xmath2 edges have lines right next to them . .
the upper panel shows the model folded through the instrument response and compared with the data .
bottom panel shows the unfolded model.,width=321 ] epic background spectra can occasionally have fluorescence emission lines of zn , cu and ni at energies above @xmath33 kev due to interaction of high energy particle with the structure surrounding the detectors and the detectors themselves ( lumb et al .
these background emission - lines vary across the ccds and are not accounted for in the standard reduction process .
we have verified that this instrumental fluorescence radiation does not affect the source or background regions we use .
nevertheless , on other regions of the ccd these features are present and can lead to false identification of narrow absorption lines in the source spectrum due to the background subtraction .
we also observe the lines at 2.68 kev and 1.47 kev claimed by pounds et al .
( 2003a ) to be from s and mg , only we identify them as different lines at much lower velocities .
the 2.68 kev line is identified here as he@xmath34 and the 1.47 kev line is identified as he@xmath34 . the combined rgs spectrum ( rgs1 and rgs2 ) of pg1211 + 143 is presented in figure [ pgtrgs ] .
numerous absorption lines and several emission lines are detected .
table [ abslines ] lists the absorption lines we identify with high certainty . for each line
we list the wavelength at the source rest - frame ( @xmath35 ) , its equivalent width ( ew ) , the line identification , and the velocity blueshift .
the absorption lines were measured by fitting each line with a local continuum and a gaussian .
we also list the identification of the lines by pounds et al .
( 2003a ) where available ( see discussion below ) .
we identify k - shell lines of c , n , o and mg and l - shell lines of o , mg , si , ar , and fe . the absorption line widths are consistent with the rgs resolution , and with the present s /
n we are not able to resolve the intrinsic velocity widths . in emission , we identify significantly broadened lines of ly@xmath2 , ly@xmath2 , the forbidden line of and its he@xmath2 resonance line , the forbidden line of , and the he@xmath2 resonance line , all in the rest frame of the source with no velocity shift . cclc|cclc @xmath36 & @xmath37 & he@xmath34 ( 7.851 ) & @xmath38 & @xmath39 & @xmath40 & ly@xmath2 ( 8.421 ) & @xmath41 + @xmath42 & @xmath43 & ly@xmath2 ( 8.421 ) & @xmath44 & + @xmath45 & @xmath46 & he@xmath2 ( 9.169 ) & & + @xmath47 & @xmath48 & blend at @xmath49 & @xmath50 & @xmath51 & @xmath52 & ly@xmath2 ( 12.134 ) & @xmath53 + @xmath54 & @xmath55 & blend at @xmath56 & @xmath57 & + & @xmath58 & @xmath59 & he@xmath2 ( 13.447 ) & @xmath60 + @xmath61 & @xmath62 & ( 15.014 ) & @xmath63 & @xmath64 & @xmath65 & ly@xmath34 ( 16.006 ) & @xmath66 + @xmath67 & @xmath68 & blend at @xmath69 & @xmath70 & + @xmath71 & @xmath72 & he@xmath73 ( 17.396 ) & @xmath74 & @xmath75 & @xmath76 & he@xmath34 ( 18.627 ) & @xmath77 + @xmath78 & @xmath79 & he@xmath80 ( 17.768 ) & @xmath81 & @xmath82 & @xmath83 & ly@xmath2 ( 18.969 ) & @xmath84 + @xmath85 & @xmath86 & he@xmath34 ( 18.627 ) & @xmath87 & + @xmath88 & @xmath89 & ly@xmath2 ( 18.969 ) & @xmath90 & + & @xmath91 & @xmath92 & he@xmath2 ( 21.602 ) & @xmath93 + @xmath94&@xmath95 & k@xmath2 ( 22.729 , 22.777)&@xmath96 & + @xmath97&@xmath98 & k@xmath2 ( 23.302 , 23.301 , 23.3)&@xmath99 & @xmath100 & @xmath101 & ly@xmath2 ( 24.781 ) & @xmath102 + @xmath103&@xmath104 & ( 27.846 , 27.881)&@xmath105 & + @xmath106 & @xmath107 & ( 28.780 ) & @xmath108 & + @xmath109 & @xmath110 & ( 29.209 ) & @xmath111 & + @xmath112&@xmath113 & ( 31.018 , 31.027)&@xmath114 & + @xmath115 & @xmath116 & ( 31.374 ) & & @xmath117 & @xmath118 & ly@xmath2 ( 33.736 ) & @xmath119 + @xmath120 & @xmath121 & ly@xmath2 ( 33.736 ) & @xmath122 & in order to quantitatively explore the emission and absorption lines , we have constructed a model for the entire rgs spectrum . the present method is an ion - by - ion fit to the data similar to the approach used in sako et al .
( 2001 ) and in behar et al .
we first use the continuum measured from the epic - pn data , but renormalized to the rgs flux level .
this continuum is then absorbed using the full set of lines for each individual ion .
our absorption model includes the first 10 resonance lines of h- and he- like ions of c , n , o , ne , and mg as well as edges for these ions .
the model also includes our own calculation for the l - shell absorption lines of fe ( behar et al .
2001 ) as well as of si , s , and ar corrected according to laboratory measurements ( lepson et al .
2003 , 2005 ) .
finally , we include inner - shell k@xmath2 absorption lines of o and mg ( behar & netzer 2002 ) , which we detect in the spectrum .
the absorbed spectrum is complemented by the emission lines mentioned above , which are observed in the rgs spectrum . by experimenting with the absorption line parameters , we find that the observed lines are all blueshifted by about 3000 ^-1 kms@xmath0 with an uncertainty of 500 ^-1 kms@xmath0(see also table [ abslines ] ) . in the model we used a turbulence velocity of 1000 ^-1 kms@xmath0 to broaden the absorption lines .
this width includes the instrumental broadening , which as noted above , we could not separate from the intrinsic broadening .
since the lines appear to be saturated , but no line goes to zero intensity in the trough , we obtain the best fit by assuming a covering factor of 0.7 for the x - ray continuum source .
the best - fit column densities that we find for the different ions are listed in table [ coldens ] .
the emission lines are modeled using gaussians with uniform widths of @xmath123 ^-1 kms@xmath0 ( resolved , but again , including the instrumental broadening ) , with no velocity shift , and assumed to be unabsorbed . these lines at fwhm @xmath124 @xmath125 ^-1 kms@xmath0 are even broader than those observed from the broad line region in the visible band ( @xmath126 ^-1 kms@xmath0 ; kaspi et al .
the model emission line fluxes are given in table [ emisflux ] .
the entire best - fit spectrum is represented in figure [ pgtrgs ] by the red curve .
the spectrum beyond 25 is particularly challenging as it comprises many unresolved lines from l - shell ions of si , s , and ar while the rgs effective area drops rapidly .
several predicted lines may be observed here ( e.g. , - 28.92 , - 30.71 , - 31.06 , - 31.93 ; these wavelengths include the 3000 ^-1 kms@xmath0 shift ) . we are still unable to explain several features seen in the data , e.g. , around 8.5 , 10.4 , or 29.8 , but the model gives a good fit to the data overall .
lcc & @xmath127 & @xmath128 + [ 0.2 cm ] & @xmath127 & @xmath129 + & @xmath130 & @xmath131 + [ 0.2 cm ] & @xmath127 & @xmath132 + & @xmath133 & @xmath134 + & @xmath135 & @xmath136 + & @xmath130 & @xmath137 + & @xmath138 & @xmath139 + & @xmath130 & @xmath137 + & @xmath140 & @xmath141 + [ 0.2 cm ] & @xmath142 & @xmath143 + & @xmath130 & @xmath132 + [ 0.2 cm ] & @xmath142 & @xmath144 + & @xmath145 & @xmath146 + & @xmath140 & @xmath147 + & @xmath133 & @xmath148 + & @xmath140 & @xmath147 + & @xmath140 & @xmath147 + [ 0.2 cm ] & @xmath127 & @xmath149 + [ 0.2 cm ] & @xmath150 & @xmath151 + [ 0.2 cm ] & @xmath135 & @xmath152 + & @xmath153 & @xmath154 + & @xmath130 & @xmath155 + [ 0.2 cm ] & @xmath133 & @xmath156 + & @xmath127 & @xmath157 + & @xmath153 & @xmath158 + & @xmath130 & @xmath159 + & @xmath140 & @xmath160 + & @xmath133 & @xmath156 + & @xmath140 & @xmath160 + & @xmath140 & @xmath160
pounds et al . ( 2003a ) detected eight absorption lines in the same rgs spectrum ( see their table 2 ) .
we list their detections ( including one line from their table 1 ) in table [ abslines ] .
the ew and wavelengths given by pounds et al .
( 2003a ) are presumably from their globally fitted model , thus there are certain discrepancies between them and our _ measured _ quantities .
we note again the small wavelength shift of about 0.1 between the data shown here and the pounds et al .
( 2003a ) data .
this does not affect our main results as it can add only about 1500 ^-1 kms@xmath0 ( at @xmath161 ) to the outflow velocity we find , which can not explain the order of magnitude difference in the outflow velocity .
we note that the velocities obtained in pounds et al .
( 2003a ) from the epic data ( their table 1 ) are in err .
these results have been revised in the erratum by pounds et al .
( 2005 ) . even in pounds
( 2005 ) the velocities span a wide range from 22600 ^-1 kms@xmath0 to 32700 ^-1 kms@xmath0 , which is weakly consistent with the 24000 ^-1 kms@xmath0 velocity claimed from the rgs data ( pounds et al .
attempting to model the data with a 24000 ^ -1 kms@xmath0 model , we were not able to reproduce all the absorption lines in table [ abslines ] . although it may be possible that both velocities are present , we suspect that most of the column density lies in the 3000 ^-1 kms@xmath0component , which we discuss below .
an outflow velocity of 3000 ^-1 kms@xmath0 places pg1211 + 143 at the high - velocity end of , or even slightly faster than , typical seyfert 1 outflows .
the more interesting parameters of the outflow such as its location and the associated mass outflow rate are much harder to obtain by means of mere absorption spectroscopy . for an extended ionization cone geometry
, the distance of the base of the cone from the continuum source for a given ionization species formed at an ionization parameter @xmath162 can be roughly estimated by ( e.g. , behar et al .
2003 ) : @xmath163 @xmath164 the standard hydrogen to electron density ratio @xmath165 in a fully ionized astrophysical plasma is 0.83 . with an ionizing luminosity , @xmath166 , of a few times 10@xmath167 erg s@xmath0 ,
this places the typical ionization component of the pg1211 + 143 absorber ( e.g. , @xmath168ergs@xmath0 cm for ) on large scales , much further out than the estimates of pounds et al .
( 2003a ) who invoke an opaque photosphere ( king & pounds 2003 ) on scales of @xmath169 cm .
lc ly@xmath2 ( 33.737 ) & 8.8 + ly@xmath2 ( 24.782 ) & 4.2 + forbidden ( 22.101 ) & 8.3 + he@xmath2 ( 21.602 ) & 3.2 + ly@xmath2 ( 18.969 ) & 5.9 + forbidden ( 13.699 ) & 2.0 + ly@xmath2 ( 12.134 ) & 1.3 + he@xmath2 ( 9.169 ) & 1.3 the total mass outflow rate of the wind is impossible to estimate without knowing the opening angle , @xmath170 ( sky covering fraction ) , of the wind .
one way to obtain @xmath170 is to measure the volume emission measure of narrow emission lines assumed to emerge from the absorbing outflow ( e.g. , behar et al . 2003 ) . in the present spectrum ,
however , we do not detect any narrow emission lines , but only broad lines , which probably originate in the inner blr .
the apparent absence of narrow emission lines could imply that the opening angle is rather small and perhaps so is the mass outflow rate
. pounds et al .
( 2003a ) obtain much higher column densities @xmath1 10@xmath171 @xmath22 than we do and their spherically symmetric expanding photosphere results in a high estimate of the mass outflow rate : @xmath172 yr@xmath0 .
mckernan et al . ( 2004 ) suggest that some portion of the absorption detected by pounds et al .
( 2003a ) could be due to hot local intergalactic medium ( igm ) since the 24000 ^-1 kms@xmath0 outflow velocity coincides with the recession velocity of pg1211 + 143 .
our interpretation , of a 3000 ^-1 kms@xmath0 outflow , is not consistent with the igm interpretation .
also , the high column density inferred , as well as the high ionization level observed , are not typical of the igm . finally no local ionized absorber is detected in the uv towards this source .
although the present model does not produce a perfect fit to the observed spectrum , it does explain self consistently the absorption lines with a single outflow velocity of 3000 ^-1 kms@xmath0 .
we stress that our method is not a global fit and that all of the ions appearing in table [ coldens ] have been identified directly in the spectrum .
the inclusion of emission lines in our model explains why some ly@xmath34-like and he@xmath34-like lines appear to be deeper than their corresponding ( saturated ) ly@xmath2 and he@xmath2 lines .
this happens due to filling of the absorption trough by the emission .
for example , the he@xmath2 absorption line is filled by emission which causes the he@xmath34 absorption line to appear deeper ( see figure [ pgtrgs ] ) . as specified in table [ coldens ] , the best - fit ionic column densities are between 2@xmath17310@xmath174 and 10@xmath175 . for o and mg , for which both k - shell and l - shell ions
are present , the l - shell ions are somewhat less abundant in the x - ray absorbing plasma , as expected .
we estimate the measured column densities to be accurate to within a factor of a few . assuming that the absorption by each ionic species is produced where it attains its maximal fractional abundance , we estimate the effective hydrogen column density , @xmath176 , by ascribing crude , peak ionic fractional abundances of 0.8 and 0.3 to k - shell and l - shell ions , respectively ( e.g. , kallman et al . 2004 ) , and by assuming solar abundances of ( 2.45 , 1.12 , 4.90 , 1.23 , 0.380 , 0.355 , 0.162 , 0.0363 and 0.320)@xmath177 for c ,
n , o , ne , mg , si , s , ar , and fe , respectively .
this yields inferred @xmath178 values for each element that are consistent to an order of magnitude .
discrepancies between elements may reflect deviations from solar abundances .
we find that the ionic column densities in our model correspond to a hydrogen column of roughly @xmath17910@xmath180 ( table [ coldens ] ) , similar to typical values measured for seyfert galaxies .
this hydrogen column density is two orders of magnitude smaller than the pounds et al .
( 2003a ) result .
we believe the detected span of charge states is primary a density effect rather than a radial stratification effect , due to the uniform outflow velocity and the similar inferred hydrogen column density , @xmath178 , for all the ions .
the observed ionic species reflect a broad and rather flat ionization distribution along the line of sight , again , an effect which has been observed in many seyfert 1 outflows ( e.g. , kaspi et al .
2001 ) as well as in seyfert 2 galaxies ( kinkhabwala et al .
in the latter case , all charge states appeared to be present at similar distances from the central source , which implies local density gradients of 34 orders of magnitude . in this sense and to the end that our diagnostics of a 3000 ^-1 kms@xmath0 outflow is correct , pg1211 + 143 resembles a seyfert 1 galaxy , only the wind is slightly faster , perhaps a consequence of its higher luminosity .
tumlinson et al .
( 2005 ) study the uv spectrum of pg1211 + 143 and find two absorption systems blueshifted by @xmath181 and @xmath182 ^-1 kms@xmath0relative to the rest frame .
they attribute them to igm in connection with two galaxy groups on the line of sight to pg1211 + 143 .
although the absorption is identified also in the uv lines the column densities they measure are too low for an x - ray detection of these igm components .
we examined the _
far ultra violet explorer ( fuse ) _ observation of pg1211 + 143 from the _ fuse _ archive taken 14 months before the _ xmm - newton _ observation .
we find no evidence for absorption in at the outflow velocities of @xmath183 ^-1 kms@xmath0 or @xmath184
^-1 kms@xmath0 . with the ionic column density from our rgs model such lines should have been detected .
the absence of the absorption in the _ fuse _ spectrum could be because of changes in the absorber between the two observations .
an alternative explanation is that the strong k@xmath2 line at 21.79 required by our model is confused with the galactic k@xmath2 absorption line which is expected at 21.76 in the rest - frame wavelength scale of figure [ pgtrgs ] .
the emission lines in the _ fuse _ spectrum appear to have a narrower component with fwhm of about 2000 ^-1 kms@xmath0 , which correspond to the width of the balmer and the emission lines ( e.g. , baskin & laor 2005 ) but also a broader component consistent with the fwhm of 6000 ^-1 kms@xmath0we find for the x - ray emission lines .
the observed fe - k absorption edge can be understood in the context of the ionized outflow .
absorption spectra of fe ions feature edges due to ejection of k - shell electrons .
for to these edges cover the range from 7.1 to 9 kev . in the more ionized cases , the spectral structure is not a simple edge as demonstrated most recently by kallman et al .
however , for ions close to neutral , an edge is an excellent approximation .
the edge energy we find in the epic - pn spectrum ( figure [ pnmodel ] ) falls at @xmath30 kev in the rest frame of pg1211 + 143 .
this corresponds to charge states of to .
if the fe edge is also due to the same 3000 ^-1 kms@xmath0 outflow observed with the rgs then it is associated with even less ionized gas .
the k - edge position of the low charge states of fe all fall close to each other and certainly can not be resolved with the current ccd instruments .
in addition , the broad ionization distribution in the outflow suggests that the observed edge is due to several charge states .
indeed , the high optical depth at the edge ( 0.56 ) requires the contribution of several ( 510 ) fe ions , with a column density of about 10@xmath175 each .
the more highly ionized fe ions ( table [ coldens ] ) also must contribute to the photoelectric absorption at higher energies .
we note that even deeper edges have been measured for similar sources ( e.g. , boller et al .
1 . we have provided a self consistent model to the ionized outflow of pg1211 + 143 revealing an outflow velocity of approximately 3000 ^-1 kms@xmath0 .
our model reproduces many absorption lines in the rgs band , although the s / n of the present data set is rather poor and some of the noise might be confused with absorption lines . 2 .
the present approach is distinct from the commonly used global fitting methods and also from the line - by - line approach used by pounds et al .
( 2003a ) .
it allows for a physically consistent fit to the spectrum and is particularly appropriate for a broad - ionization - distribution absorber as observed here for pg1211 + 143 .
the present model also features several broad ( fwhm = 6000 ^-1 kms@xmath0 ) emission lines , which are observed directly in the data .
a broad and relatively flat ionization distribution is found throughout the x - ray outflow consistent with a hydrogen column density of roughly @xmath179@xmath185 .
this is reminiscent of the outflow parameters measured in other well studied seyfert galaxies .
we also detect fe - k absorption , which was identified by pounds et al .
( 2003a ) as a strongly blueshifted absorption line .
we find that most of the fe - k opacity can alternatively be attributed to several consecutive , low charge states of fe , although it can not be assessed whether the absorber is co - moving with the outflow or not .
future missions with microcalorimeter spectrometers on board might be able to address this interesting question .
we thank d. chelouche and a. laor for helpful discussions .
we are grateful for several valuable suggestions by an anonymous referee .
this research was supported by the israel science foundation ( grant no .
28/03 ) , and by a zeff fellowship to s.k .
behar , e. , rasmussen , a. p. , blustin , a. j. , sako , m. , kahn , s. m. , kaastra , j. s. , branduardi - raymont , g. , & steenbrugge , k.c .
2003 , , 598 , 232 boller , th . , tanaka , y. , fabian , a. , brandt , w. n. , gallo , l. , anabuki , n. , haba , y. , & vaughan , s. 2003 , , 343 , l89 chartas , g. , brandt , w. n. , gallagher , s. c. , & garmire , g. p. 2002
, , 579 , 169 chartas , g. , brandt , w. n. , & gallagher , s. c. 2003 , , 595 , 85 crenshaw , d. m. , kraemer , s. b. , & george , i. m. 2003 , , 41 , 117 pounds , k. a. , reeves , j. n. , king , a. r. , page , k. l. , obrien , p. t. , & turner , m. j. l. 2003a , , 345 , 705 pounds , k. a. , king , a. r. , page , k. l. , & obrien , p. t. 2003b , , 346 , 1025 pounds , k. a. , reeves , j. n. , king , a. r. , & page , k. l. 2004 , , 350 , 10 pounds , k. a. , reeves , j. n. , king , a. r. , page , k. l. , obrien , p. t. , & turner , m. j. l. 2005 , , 356 , 1599 reeves , j. n. , obrien , p. t. , & ward , m. j. 2003 , , 593 , 65 sako , m. , et al . 2001 , , 365 , l168 | we analyze the x - ray spectrum of the quasar pg1211 + 143 observed with the ccd and grating spectrometers on board _ xmm - newton_. using an ion by ion fitting model we find an outflow component of about 3000 ^-1 kms@xmath0that includes absorption lines of k - shell and l - shell ions of the astrophysically abundant elements .
we also identify and include in our model broad ( fwhm = 6000 ^ -1 kms@xmath0 ) emission lines from h - like ions of c , n , o , and ne , and he - like ions of o , ne , and mg .
the outflow velocity we find is an alternative interpretation of the data and is in contrast with the ultra high velocity of @xmath1 24000 ^-1 kms@xmath0 reported previously for this object .
nevertheless , we can not completely rule out the presence of a high velocity component due to the poor signal - to - noise ratio of the data . |
the emergence of class structure in society is a ubiquitous phenomenon in the biological and the social sciences @xcite .
social hierarchies have been widely observed in animal populations including insects @xcite , mammals @xcite , and primates @xcite , as well as human communities @xcite . the possibility of quantitative modeling of social phenomena using concepts and techniques borrowed from the physical sciences is rapidly gaining appreciation .
examples of such modeling include the wealth distribution @xcite , opinion dynamics @xcite , and rumor propagation @xcite .
such approaches typically draw analogies between individual agents in the social system and particles in a corresponding physical system and then identifying macroscopically observed phenomena with microscopic agent - agent interactions @xcite . in this spirit , we seek to tie the emergence of social structures to specific interactions between agents within a general version of the recently - introduced advancement - decline process @xcite . in our model ,
the social standing of each agent is characterized by a single number , its fitness .
agents increase their fitness by interacting with other agents and also , their fitness may decline spontaneously .
this simple model has only two parameters : the probability that the fitter agent advances in an interaction and the rate of decline .
we find that a rich variety of familiar social structures emerges as a result of the competition between advancement and decline .
when decline dominates , the society is static and the fitness distribution approaches a steady state .
when the decline rate is comparable to the advancement rate , the society is dynamic and the characteristic fitness of the population increases linearly with time . in this case
, there are several possibilities . when the less fit agent benefits from social interactions , an egalitarian society arises in which all agents advance at the same rate . consequently , inequalities among agents are small . on the other hand , when the fitter agent tends to benefit in competitions , agents advance at different rates and social inequalities increase with time .
depending on the relative influence of advancement and decline , either the entire population or only a fraction of it may be upwardly mobile . in the latter case ,
the society consists of a static lower class and an upwardly - mobile middle class .
in section [ adp ] , we introduce the general advancement - decline process and the governing master equations .
the overall class structure and the statistics of the mobile middle class are obtained using scaling analysis in section [ mobility ] .
the basic features of the egalitarian society are investigated in section [ equality ] , where the cumulative fitness distribution may be largely determined by linear traveling wave analysis . in section [ poverty ] , the statistics of the lower class , where the fitness distribution is steady , are determined .
we conclude in section [ conc ] .
we model a scenario in which the social status of an agent benefits from increased social interactions , while solitude or isolation have the opposite effect . indeed
, highly connected individuals often have better access to information , resources , and power , that are often gained as a result of social interactions .
thus , in our model there are two competing evolutionary processes that influence the fitness of agents : ( i ) advancement via social interactions , and ( ii ) decline due to the lack of interactions ( fig
. [ processes ] ) . for simplicity , social standing
is represented by a single parameter , the integer - valued fitness @xmath3 . 0.3 in _ ( i ) advancement .
_ agents interact in pairs , and as a result of the interaction , only one advances .
there are two possibilities : either the fitter agent advances or the less fit advances .
we allow the fitter agent to advance with probability @xmath0 and the less fit agent to advance with probability @xmath1 .
thus , when two agents with fitness @xmath4 and fitness @xmath5 interact ( with @xmath6 ) , the outcome is @xmath7 for @xmath8 the fitter agent always advances @xcite , while for @xmath9 the less fit agent always advances .
the interaction rules are defined so that one randomly - chosen agent advances when two equally - fit agents interact . without loss of generality ,
the interaction rate is set to @xmath10 .
also , we consider the thermodynamic limit where the number of agents is infinite . _
( ii ) decline .
_ in the decline step , the fitness of an individual decreases according to @xmath11 with rate @xmath2 .
this process reflects a natural tendency for fitness to decrease in the absence of social activity .
we impose the lower limit for fitness to be @xmath12 ; once an individual reaches zero fitness , there is no further decline .
our goal is to understand how the fitness distribution of a population evolves as a function of the two model parameters , the advancement probability @xmath0 and the decline rate @xmath2 .
let @xmath13 be the fraction of agents with fitness @xmath4 at time @xmath14 . in the mean - field limit , where any pair of agents is equally likely to interact
, the fitness distribution obeys the master equation @xmath15 here @xmath16 and @xmath17 are the respective cumulative distributions of agents with fitness less than @xmath4 and fitness greater than @xmath4 .
the boundary condition is @xmath18 .
the first pair of terms accounts for decline , the second pair of terms describes interactions where the stronger agent advances , and the third pair of terms accounts for interactions where the weaker agent advances
. the last pair of terms describes interactions between two equal agents and it reflects that when two such agents interact , only one of them advances .
the prefactor @xmath10 arises because there are half as many ways to chose equal agents as there are for different agents .
we consider the initial condition where all agents have the minimal fitness @xmath19 .
it proves useful to rewrite the evolution equation in a closed form that involves only the cumulative distribution . summing the rate equations ( [ dis - eq ] ) and using the relations @xmath20 and @xmath21
, the cumulative distribution @xmath22 obeys @xmath23 the boundary conditions are @xmath24 , @xmath25 , and the initial condition is @xmath26 for @xmath27 .
there is a one - to - one correspondence between the four terms in equations ( [ dis - eq ] ) and ( [ cum - eq1 ] ) .
the master equation for the cumulative distribution can be simplified by consolidating the advancement terms @xmath28 the mean fitness @xmath29 evolves according to @xmath30 this result can be derived directly by summing the master equations ( [ dis - eq ] ) or even simpler , from the definition of the advancement - decline process .
the first term accounts for advancement , where interactions occur with rate @xmath10 such that each interaction advances only one agent .
the second term stems from decline and reflects the fact that all agents except for the least - fit ones decline with rate @xmath2 .
we now discuss the basic social structures that emerge from the solution to the master equation .
we determine the class structure of the population via a simple scaling analysis of the master equation .
let us take the continuum limit of the master equation by replacing differences with derivatives , @xmath31 . to first order in this `` spatial '' derivative , we obtain the nonlinear partial differential equation @xmath32 \frac{\partial f}{\partial k}\,.\ ] ] when the spatial derivative and the temporal derivative balance , the typical fitness increases linearly with time , @xmath33 .
therefore , we make the scaling ansatz @xmath34 the boundary conditions are @xmath35 and @xmath36 . substituting this scaling form in eq .
( [ cum - eq - cont ] ) , the partial - differential equation reduces to the ordinary differential equation @xmath37\frac{d\phi}{dx}=0,\ ] ] where the prime denotes differentiation with respect to the scaling variable @xmath38 .
the solution is either @xmath39 , _
i.e. _ , @xmath40 or the linear function @xmath41 using these two solutions and invoking ( i ) the boundary conditions @xmath35 and @xmath36 , ( ii ) the bounds @xmath42 , ( iii ) monotonicity of the cumulative distribution , @xmath43 , and ( iv ) the assumption that the scaling function changes continuously with @xmath0 and @xmath2 , we can then deduce the four possible social structures of the population .
middle - class society _ : we first examine the conditions for the linear scaling function ( [ scaling - sol - lin ] ) to apply .
first , the cumulative scaling function ( [ scaling - form ] ) must be a monotonically increasing function .
therefore , the linear solution ( [ scaling - sol - lin ] ) holds only when its slope is positive , that is , when .
second , the scaling function is bounded , ; this condition implies the lower and upper bounds @xmath44 on the scaled fitness .
the obvious constraints @xmath45 and @xmath46 lead to the conditions @xmath47 and @xmath48 . by imposing continuity , as well as the limiting behaviors @xmath49 and @xmath50 outside the linear region , the scaled cumulative distribution is the piecewise linear function ( fig .
[ fig - phi - mc ] ) : @xmath51 this behavior describes a middle class society where _ all _ agents are upwardly mobile , as their fitness improves linearly with time . in this case
, social inequalities also increase indefinitely with time : the agents at the bottom of the middle class have fitness @xmath52t$ ] and the richest agents have fitness @xmath53 .
the middle - class society lies within the triangular region defined by the lines @xmath54 , @xmath55 , and @xmath56 , shown in fig .
[ classes ] . versus @xmath57 at different times for @xmath58 and @xmath59.,scaledwidth=40.0% ]
hierarchical society _ : along the line @xmath54 , the fitness of the poorest agents vanishes .
moreover , the linear scaling solution ( [ scaling - sol - lin ] ) has a finite positive value at zero fitness for a range of parameter values @xmath0 and @xmath2 .
these two observations suggest the existence of another type of piecewise linear solution with @xmath60 .
the bounds @xmath61 impose the conditions @xmath62 and @xmath63 . in this region , the scaling function has two distinct components ( fig .
[ fig - phi - hs ] ) @xmath64 thus , we find a hierarchical society ( fig .
[ classes ] ) that includes both an upwardly - mobile middle class and a static lower class .
the lower class consists of a finite fraction @xmath65 of agents with zero fitness ( in scaled units ) . in section [ poverty ] , we examine the lower class more closely and show that its fitness distribution is time - independent and extends only over a finite range .
_ 3 . lower - class society _ : when the fraction @xmath66 of agents with zero fitness reaches 1 , the entire population is poor . for @xmath67 ,
the condition @xmath68 occurs on the boundary @xmath69 . at this point
the fitness distribution becomes a step function , @xmath70 with @xmath71 for @xmath72 and @xmath73 for @xmath74 .
we therefore conclude that there is a region of the phase diagram where the scaled fitness of the entire population is zero . for any initial state
, the fitness distribution quickly approaches the step - function in a lower - class society . versus @xmath57 at different times for @xmath75 and @xmath76.,scaledwidth=40.0% ]
_ 4 . egalitarian society _ : there is another region of the phase diagram where the fitness distribution also becomes a step function . when @xmath56 and @xmath77 , then both @xmath78 and @xmath79 are equal to @xmath80 .
therefore with @xmath81 . since the scaling function must change in a continuous fashion , we conclude that for @xmath82 , the scaling function is again a step function but with a jump at non - zero fitness .
that is @xmath83 in this egalitarian society , all agents have the same scaled fitness @xmath84 or alternatively the fitness @xmath85 .
the velocity @xmath86 follows easily from the average fitness ( [ kdot ] ) .
since all agents advance at constant rate , then the term @xmath87 is negligible and therefore , the propagation velocity is @xmath88 in section [ equality ] , we show that in this society , the fitness differences between agents are small and do not grow with time .
this is the sense in which the society is egalitarian .
when @xmath82 , the weaker agent preferentially benefits in an interaction , so that the rich effectively supports the poor .
we also note that the lower class and the egalitarian society share one common feature : they do not have a middle class .
the boundary between these two phases , determined by the condition @xmath89 , is the line @xmath76 ( fig .
[ classes ] ) .
our numerical integration of the evolution equations confirms the overall picture of four different social structures ( fig .
[ classes ] ) : a middle class society ( fig .
[ fig - phi - mc ] ) , a hierarchical society ( fig . [ fig - phi - hs ] ) , a lower - class society as in ( [ phi - lc ] ) , and an egalitarian society as in ( [ phi - es ] ) .
the numerical data was obtained by integrating @xmath22 for @xmath90 using a fourth - order adams - bashforth method @xcite , with a specified accuracy of @xmath91 in the distribution @xmath22 . our numerical results for the cumulative distribution @xmath22 , when plotted versus the scaling variable @xmath57 , smoothly approaches the appropriate expressions for the piecewise linear scaling function @xmath92 derived in the previous subsection ( figs .
[ fig - phi - mc ] & [ fig - phi - hs ] ) . as time increases
, the fitness distribution narrows .
the simulations also show that the approach to the scaling solution is slowest in the vicinity of the extremes of the middle class @xmath93 and @xmath94 ( with @xmath95 for the hierarchical phase ) .
the correction to scaling near these extrema can be determined by keeping derivatives up to second order in the continuum limit of the master equation .
this approximation gives the nonlinear diffusion equation @xcite @xmath96 + ( p-1/2)\left(\frac{\partial f}{\partial k}\right)^2.\nonumber\end{aligned}\ ] ] the linear terms are separately displayed in the first line and the nonlinear terms in the second .
let us first consider the poorest agents , _
i.e _ , the behavior close to @xmath93 . since the cumulative fitness distribution is small near this point , the nonlinear terms can be neglected and the governing equation ( [ cum - eq - second ] ) reduces to the standard convection - diffusion equation @xmath97 with propagation velocity @xmath98 and diffusion coefficient @xmath99 .
indeed , since the fitness distribution is obtained from the cumulative distribution by differentiation , @xmath100 , the fitness distribution satisfies the same equation ( [ cum - eq - cont ] ) as the cumulative distribution . for the middle - class society , we therefore conclude that the bottom of the middle class has a gaussian tail , with the center of the gaussian located at @xmath101 and with width @xmath102 .
the same analysis can be carried out for the hierarchical society , where the quantity @xmath103 now satisfies the diffusion equation with zero velocity @xmath104 and diffusivity @xmath105 .
conversely , the distribution for the top end of the middle class can be obtained by analyzing @xmath106 .
it is immediate to show that this quantity again obeys eq .
( [ de ] ) with velocity @xmath107 and diffusivity @xmath108 .
we conclude that the extremes of the middle class are characterized by gaussian tails whose extents grow diffusively with time . in terms of the scaling variable @xmath109 ,
the deviation from the scaling function @xmath92 is appreciable only within a region of whose width is shrinking as @xmath110 .
for the special case @xmath56 , the nonlinear terms vanish and the fitness distribution is described _ exactly _ by the linear convection - diffusion equation ( [ de ] ) with drift velocity @xmath81 and diffusion coefficient @xmath111 ( the nonlinear term is negligible ) . thus there is a drift toward smaller fitness for @xmath112 and the fitness distribution approaches a steady - state profile that decays exponentially with fitness . in the opposite case of @xmath113 ,
the fitness distribution is simply a gaussian that drifts to larger fitness with velocity @xmath114 and whose width is proportional to @xmath115 . in the case of @xmath56 ,
the relative position of an agent in the society is irrelevant and advancement reduces to a pure random walk @xcite .
in the egalitarian phase , the step function form of the scaling solution , eq .
( [ phi - es ] ) , suggests that the fitness distribution has the traveling wave form @xmath116 with the propagation velocity ( [ v ] ) .
this is confirmed by numerical integration of the master equation ( [ cum - eq ] ) , as shown in fig .
( [ fig - wave ] ) . to determine the shape of the wave @xmath117
analytically , we substitute the waveform ( [ wave ] ) into the master equation ( [ cum - eq ] ) to give the nonlinear difference - differential equation for @xmath117 @xmath118\!+ \!(1\!-\!p)[u(z\!-\!1\!)-\!u(z)]\nonumber\\ & + & ( p-1/2)[u^2(z-1)-u^2(z)].\end{aligned}\ ] ] the boundary conditions are @xmath119 and @xmath120 . versus @xmath121 , where @xmath81 is the speed of wavefront at different times for @xmath122 and @xmath59 ( egalitarian society ) .
[ fig - wave],scaledwidth=40.0% ] we apply standard linear analysis in the tail regions to deduce the leading and trailing shapes of the waveform .
when @xmath123 , then @xmath124 and therefore @xmath125 . to first order in @xmath126 , eq .
( [ wave - eq ] ) becomes @xmath127+(1\!-\!p)[u(z\!-\!1)-u(z)]=0.\end{aligned}\ ] ] the behavior in this case is determined by the balance between decline and advancement events where the less fit agent advances .
the solution to this linearized equation is the exponential decay @xmath128 substituting this form and ( [ v ] ) into the linearized equation , the decay constant @xmath129 is the root of the following equation @xmath130.\ ] ] similarly , in the limit @xmath131 we linearize the wave equation ( [ wave - eq ] ) for the small quantity @xmath132 to obtain @xmath133+p[r(z)-r(z\!-\!1)].\end{aligned}\ ] ] in this case the behavior at large fitness is governed by the balance between decline and advancement events where the fitter agent advances @xcite .
the solution to the above differential equation is again the exponential decay @xmath134 with the decay constant @xmath135 satisfying @xmath136.\ ] ] we conclude that the likelihood of having agents that are much richer or much poorer than the average fitness @xmath137 in the egalitarian society is exponentially small , as illustrated in fig .
[ fig - tails ] .
the society therefore consists of agents whose fitnesses are all roughly the same , @xmath85 .
as one might naturally anticipate , social inequalities are small under the dynamics in which the rich preferentially gives to the poor . versus @xmath121 for @xmath138 .
the theoretical predictions ( [ wave - poor])([v - beta ] ) with @xmath139 and @xmath140 are indicated by straight lines.,scaledwidth=40.0% ] for the case where the less fit agent always advances , the fitness distribution has a special form . in this case
, the complementary cumulative distribution obeys @xmath141 with the initial condition @xmath142 and the boundary condition @xmath143 .
we expect that the fitness distribution will continue to have the form of a propagating wave .
substituting the traveling wave form @xmath144 into the master equation ( [ cum - eq2 ] ) gives @xmath145+\frac{1}{2}\left[r^2(z-1)-r^2(z)\right].\ ] ] an exponential solution does not give asymptotic balance of terms as @xmath131 , and we therefore attempt a solution of the form @xmath146 . substituting this form into the above equation and keeping only the dominant term @xmath147 on the right - hand - side gives @xmath148 for the positive terms on the left and the right hand side to balance , the dominant exponential terms must first balance , yielding the recursion equation @xmath149 .
the solution is the exponential @xmath150 .
balancing the prefactors , @xmath151 yields @xmath152 . as a result ,
the decay in the tail region is super - exponential @xcite @xmath153 as @xmath131 .
the constants @xmath154 and @xmath86 should be determined numerically .
hence , the front of the traveling wave is extremely sharp .
this tail characterizes statistics of the rich , so when the rich never benefits from interactions with the poor , rich agents are ultra - rare ( fig .
[ fig - se ] ) .
even though the leading tail extends to only a handful of sites , it is still possible to verify the super - exponential decay ( [ super - exp ] ) . in contrast , the @xmath155 tail that characterizes the poor is not altered ; it has the same exponential tail as in eq .
( [ wave - poor ] ) . versus @xmath156 , obtained by numerical integration of ( [ cum - eq2 ] ) to time @xmath157 with @xmath55 .
, scaledwidth=40.0% ]
we now determine the fitness distribution of poorest agents , a class that exists in both the hierarchical and lower - class societies . as we shall now show , the fitness distribution of the lower class in the limit of small but non - zero fitness approaches a steady state .
for the case of the hierarchical society , we write @xmath158 , where @xmath66 is the lower - class fraction , with the deviation @xmath159 vanishing for large @xmath4 .
substituting this form into the master equation ( [ cum - eq ] ) and setting the time derivative to zero , gives @xmath160 consider first the lower - class society , for which the entire population is poor , @xmath68 .
using this fact , and neglecting terms of the order @xmath161 , we find @xmath162 the solution to this equation is simply the exponential form @xmath163 with @xmath164 .
since @xmath165 , then @xmath166 in the lower - class society , the fitness is confined to a very small range .
notice also that this exponential decay coincides with the traveling wave solution ( [ wave - rich ] ) with @xmath86 set equal to zero , as the decay function is now @xmath167 . finally , we consider the hierarchical society . using ( [ l ] ) and following the same steps that led to eq .
( [ g - lc ] ) , we obtain @xmath168 with @xmath169 . to determine @xmath159 , we expand the differences to second order and
assume that @xmath170 to give , after straightforward steps , @xmath171 .
the asymptotic solution to this equation is @xmath172 . finally , using @xmath20 and @xmath158
, we determine the fitness distribution from @xmath173 to be @xmath174 thus , for the hierarchical society , the fitness distribution has a power - law large - fitness tail in the lower class region ( see also @xcite for more details ) . as discussed in section [ mobility ] , there is a diffusive boundary layer that separates the steady - state fitness distribution in the lower class and time dependent fitness distribution in the middle class . from eqs .
( [ scaling - form ] ) and ( [ phi - hs ] ) , the fitness distribution in the middle class is @xmath175^{-1}$ ] . equating this expression with eq .
( [ fk - powerlaw ] ) gives a crossover scale @xmath176 thus , the steady - state region extends over a fitness range that grows as @xmath177 .
we also note that this crossover scale agrees with the diffusivity @xmath105 , obtained in section [ mobility ] . in terms of the variable @xmath57
, the size of this region @xmath178 decays with time .
thus , a diffusive boundary layer separates the lower class and the middle class .
in summary , we have seen that the competition between advancement and decline leads to a rich and realistic set of possible social structures . from the master equation for the underlying fitness distribution ,
we obtain three types of classes : a static lower class , a mobile but disperse middle class , and a mobile but `` condensed '' egalitarian class .
the population as a whole organizes into four types of societies , three of which consist of one of these classes , and a hierarchical society in which the lower class and the middle class coexist . two parameters , the rate of decline and the advancement probability , quantify the competition between advancement and decline .
the overall social organization is determined solely by these two parameters .
the fitness distribution has a very different character in each of the classes . in the lower class ,
this fitness distribution approaches a steady state . in the middle class ,
the distribution is self - similar in time and correspondingly the characteristic fitness increases linearly with time .
although agents are upwardly mobile , the disparities between agents in the middle class also grows indefinitely . in the egalitarian class ,
the fitness distribution follows a traveling wave , so that all agents constantly advance , but fitness differences between agents remains small . much of the richness of the phenomenology is due to the fact that the mechanisms for advancement and decline are fundamentally different .
one requires interaction between agents , while the other is a single - agent process .
this dichotomy is reflected by the master equation where the decline terms are linear but the advancement terms are nonlinear . as a result , there is no detailed balance and the dynamics are non - equilibrium in character
. it should be interesting to use the advancement - decline model to analyze real - world data .
one natural application is to wealth and income distributions of individuals , where both power - law and exponential behavior has been observed @xcite .
a related issue is the wealth of nations .
it is well documented that the wealth distribution of countries is extremely inequitable , with 60% of the world s population producing just 5.6% of the planet s gross domestic product ( gdp ) , another 20% producing 11.7% , and the remaining 20% of the population producing 82.7% of the gdp @xcite .
the existence of such a large underclass corresponds to a large decline rate in our diversity model and it may be worthwhile to understand the social mechanisms for such a large decline .
another possibility is sports statistics where the winning percentage distribution of teams plays the role of the fitness distribution @xcite .
m. girvan and m. e. j. newman , proc .
usa * 99 * , 7821 ( 2002 ) .
w. weidlich , _ sociodynamics : a systematic approach to mathematical modelling in the social sciences _ ( harwood academic publishers , 2000 ) | we model the dynamics of social structure by a simple interacting particle system .
the social standing of an individual agent is represented by an integer - valued fitness that changes via two offsetting processes .
when two agents interact one advances : the fitter with probability @xmath0 and the less fit with probability @xmath1 .
the fitness of an agent may also decline with rate @xmath2 . from a scaling analysis of the underlying master equations for the fitness distribution of the population , we find four distinct social structures as a function of the governing parameters @xmath0 and @xmath2 .
these include : ( i ) a static lower - class society where all agents have finite fitness ; ( ii ) an upwardly - mobile middle - class society ; ( iii ) a hierarchical society where a finite fraction of the population belongs to a middle class and a complementary fraction to the lower class ; ( iv ) an egalitarian society where all agents are upwardly mobile and have nearly the same fitness .
we determine the basic features of the fitness distributions in these four phases . |
looking out at the universe from the vantage point of earth , we see a small fluctuation @xmath0 in the temperature of the cosmic microwave background radiation ( cmb ) . over the last decade , advances in observational astronomy have allowed us to commence a detailed study of @xmath0 s statistical properties .
we aim to compare these properties with those of small fluctuations which , according to our present ideas , are expected to have existed in the very early universe .
these fluctuations evolve in a hot , dense plasma of tightly coupled baryons and photons .
if we suppose the plasma era was preceded by an epoch of primordial inflation , then initial conditions for these fluctuations can be calculated from the parameters of the inflationary model . the variance predicted by this approach
was computed in a now - classic series of early papers @xcite , and later extended to the skewness @xcite and kurtosis @xcite . with sufficiently precise experimental data in hand
, we can work backwards from observations of these statistical properties to determine what the initial conditions must have been .
the result is a realistic prospect of constraining parameters in certain inflationary models .
this ambitious programme was begun in earnest following the arrival of high - quality cmb maps made by the _ wilkinson microwave anisotropy probe _ ( wmap ) @xcite .
even more precise data are expected from the _ planck surveyor _ satellite .
once it had become clear that the statistical properties of @xmath0 carried important information about whatever physics was operative during inflation , theorists were soon tempted to refine their calculations in the hope that additional detail could be extracted .
a similar programme in electroweak physics had provided important clues about the unknown details of electroweak symmetry breaking , and it was reasonable to explore the possibility of similar gains in cosmology .
the observable statistical properties of the inflationary density perturbation are encoded in its @xmath1-point correlation functions , with the leading contribution to each function typically suppressed by @xmath2 powers of the small quantity @xmath3 @xcite , where @xmath4 is the hubble rate during inflation and @xmath5 is the planck mass .
two types of refinement were possible : either to calculate to lowest order in @xmath3 at progressively larger @xmath1 , or to calculate higher order corrections with @xmath1 fixed .
it had been known for a long time that perturbation theory beyond leading order in @xmath3 was complicated by troublesome infrared behaviour .
sasaki , suzuki , yamamoto & yokoyama considered a @xmath6 theory with @xmath7 , coupled non - minimally to gravity , and evaluated each @xmath1-point function perturbatively @xcite .
noticed that the @xmath1-point functions evaluated in de sitter space had a potential divergence in the far future , which they called `` superexpansionary . ''
we will rephrase the analysis of sasaki _ et al . _ in modern notation and discuss the meaning of these divergences in [ sec : zoo - time ] and [ sec : time ] .
later , mukhanov , abramo & brandenberger studied a different sort of infrared effect the accumulation of long wavelength fluctuations in an expanding universe @xcite .
( see also abramo & woodard , ref .
this differed from the analysis of sasaki _ et al .
_ , which had been based on a purely geometrical effect .
whether such divergences are physical is complicated by the question of gauge , initially studied by unruh @xcite .
later , losic & unruh were able to give a gauge - invariant argument that infrared terms generically become large in spacetimes close to de sitter @xcite
. we will discuss corrections of this type in [ sec : zoo - box ] and [ sec : space ] .
divergences of a third kind had long been studied by prokopec , tsamis , woodard and their collaborators .
these authors took the quantum nature of correlation functions seriously , and calculated loop corrections just as one would when studying scattering experiments .
unfortunately , the complexity of both the rules for calculating these loops and the computations themselves are somewhat greater than in minkowski space .
these calculations have been reported in a large literature .
together with quantum effects studied by others , including boyanovsky , de vega & sanchez , a recent sample can be found in refs .
we will consider effects of this type in [ sec : zoo - new ] .
perturbative calculations of correlation functions from inflation had been studied in the 1980s by allen , grinstein & wise @xcite and later by falk , rangarajan & srednicki @xcite .
the sophistication of such calculations increased dramatically in the years following maldacena s calculation of the three - point function in single - field slow - roll inflation @xcite . at least in part , this increase was a consequence of the more complicated models of inflation which had been developed during the 1990s and early 2000s , and which theorists hoped would yield a distinctive pattern of correlations @xcite .
these developments are reviewed pedagogically in a recent article by chen @xcite . since 2002 ,
significant theoretical effort has been expended in refining our understanding of both inflationary correlation functions themselves , and the models which can generate significantly non - gaussian statistics .
the calculations reviewed in this paper were a consequence of the reinvigoration of interest in infrared issues which followed this effort .
logarithms exhibiting secular growth with time were discovered in calculations of @xmath1-point correlation functions with @xmath7 @xcite .
other calculations , if taken at face value , gave divergent answers in the infinite - volume limit .
divergences could be avoided by carrying out the calculation in a finite box , but any such procedure left behind residual logarithms involving the box cutoff @xcite .
further logarithms were encountered while studying the influence of unknown physics at energy scales even larger than the inflationary scale @xcite .
infrared effects are not a phenomenon unique to de sitter calculations , but are present in many applications of quantum field theory .
their appearance typically signals the presence of a non - trivial background , which can not be described by asymptotic _ in_- and _ out_-states containing a definite number of particles . in scattering calculations ,
infrared divergences allow putative fixed - particle states to develop a slowly varying field , made out of an accumulation of arbitrarily many soft particles radiated on approach to or recoil from the scattering event .
this `` bloch nordsieck '' or `` initial state radiation '' phenomenon is studied in many textbooks on quantum field theory @xcite .
a similar effect can occur in the confined phase of quantum chromodynamics , where soft particles can be radiated by partons moving inside nuclei .
these particles are trapped by the strong qcd force , and change the background colour field .
this effect may dramatically alter the mix of partons observed by an impinging probe such as a photon , and can not be neglected in an accurate comparison with experiment @xcite .
the effective @xmath8 approximation is a related example , in which the same infrared effects control the @xmath9 content of colliding nuclei .
the results are very practical .
for example , this method can be used to predict the rate of higgs production from @xmath10 fusion @xcite . throughout this article , planck s constant @xmath11 and the speed of light @xmath12
are set to unity .
many expressions are written in terms of the reduced planck mass , @xmath13 .
for brevity , a scalar field which contributes to the energy
momentum tensor @xmath14 is referred to as ` active , ' whereas a field making no contribution to @xmath14 is referred to as a ` spectator . ' in a cosmological context , such scalars are isocurvature modes .
in this section , i briefly review the sources of logarithmic divergences in inflationary correlation functions . a decade
after calculations of the 2-point function had determined the variance in @xmath0 expected from an inflationary initial condition , the corresponding skewness was calculated by falk , rangarajan & srednicki @xcite in an approximation where gravitational interactions were neglected . at that time
, data from the cobe satellite @xcite had established the existence of temperature fluctuations in the microwave background at the level @xmath15 , where @xmath16 was the average microwave temperature over the sky . in inflationary models @xmath17
is closely related to the parameter @xmath3 . since the skewness is proportional to @xmath18 , it was already evident that any non - gaussianity would be very small .
let us suppose inflation is driven by a single scalar field with potential @xmath19 .
we define the slow - roll parameters by @xmath20 and use a conformal time variable , @xmath21 , related to cosmic time @xmath22 by @xmath23 . when measured in this variable
, horizon crossing occurs for a mode of wavelength @xmath24 when @xmath25 , and after @xmath26 e - folds outside the horizon we have @xmath27 .
the infinite future of de sitter space corresponds to @xmath28 . in this notation
the result of falk , rangarajan & srednicki can be written @xmath29 where a subscript ` @xmath30 ' denotes evaluation at the time @xmath31 , and ` @xmath32 ' denotes terms of lower order in the slow - roll expansion , together with other terms which do not grow as @xmath33 @xcite . , which is conserved on superhorizon scales , @xmath34 , in the absence of entropy modes .
occasionally it has been argued that an estimate of the non - gaussian yield at the end of inflation can be obtained by setting @xmath35 , but this example shows clearly that unless care is taken to account for possible cancellations any such procedure can be quite misleading .
the possibility of such cancellations when reasoning with cutoff - dependent terms in a general effective field theory was pointed out by burgess & london @xcite . in the inflationary context
, the pitfalls of this procedure were noted in ref .
i would like to thank xingang chen for discussions on this point . ]
higher powers of @xmath36 may be generated at higher order in the slow - roll expansion , either by retaining such terms explicitly in the tree - level calculation or by including loop corrections , as we shall shortly explain .
a similar effect can be observed in the massless limit of the calculation by chen & wang @xcite .
the result is a power series in @xmath36 with an apparent divergence in the limit @xmath37 , as sasaki _
et al . _ had predicted @xcite .
series representations of this type were later applied to inflationary correlation functions by gong & stewart @xcite , who obtained them by systematically solving mukhanov s equation @xcite using a green s function approach .
gong & stewart remarked that the correction linear in logarithms was typically proportional to @xmath38 , whereas the correction quadratic in logarithms was typically proportional to @xmath39 , and so on .
always orders itself in this form , and according to a general theorem due to weinberg ( to be discussed in
[ sec : weinberg ] ) there may be cases where it does not .
in many practical examples where the slow - roll approximation applies , however , this structure appears . in these expressions , the symbol ` @xmath40 ' is used to mean that the coefficient of @xmath41 is a quantity of order @xmath42 ( but not necessarily @xmath42 itself ) , wheras the coefficient of @xmath43 is of order @xmath44 , and so on .
] one could therefore expect the series to break down as a predictive instrument when @xmath45 .
this occurs when the mode @xmath24 is of order @xmath46 e - folds outside the horizon .
taking into account the construction of gong & stewart , and bearing in mind that @xmath31 is the time at which we wish to evaluate the correlation functions , it seems clear that there is nothing mysterious about the appearance of powers of @xmath36 .
such terms merely express that correlations evolve outside the horizon .
this evolution is mainly driven by the evolving classical background field @xcite , but we shall see in [ sec : dynamical ] that intrinsically quantum contributions can also be present . we conclude that , as in other applications of quantum field theory , properties of the background play a role in interpreting the physics of infrared divergences . in the present case ,
correlation functions can be evaluated at any time of interest provided the series of logarithms can be summed or ignored .
therefore , for observational purposes the key question is the value of @xmath42 when modes of interest leave the horizon . we would typically wish to evaluate correlation functions at the end of inflation , where initial conditions for the subsequent evolution must be set . if @xmath47 , the end of inflation may be uncomfortably close to the era when @xmath48 .
we will return to this question in [ sec : time ] , where we will argue that a reasonable prescription exists for handling time - dependent logarithms which does not impair our ability to extract a predictive initial condition even if @xmath49 at the time of interest .
we can already note one obvious strategy .
if only logarithmic divergences are present , then better control over the power series may be achieved by `` resumming'' that is , accounting for the contribution of all terms at the same order as the _ leading logarithms _ , which are terms of the form @xmath50 for all @xmath51 @xcite . for this purpose
the principal tool is the renormalization group equation .
such an analysis has been carried out by burgess , holman , leblond & shandera @xcite , whose method will be outlined in [ sec : time ] .
unfortunately , this method leaves open the question of whether all divergences in the @xmath33 limit are comparatively tame logarithms , or if more aggressive behaviour can occur .
this question was taken up by weinberg in 2005 @xcite , and later by chaicherdsakul @xcite .
weinberg was able to prove that in many inflationary models , but not all , the worst divergences would be logarithmic .
in a later publication , this theorem was extended to fields of higher spin @xcite .
what of the possibility of faster divergences , which are apparently allowed by the analysis of sasaki _ et al . _
these would grow like powers of the scale factor , @xmath52 , corresponding to powers rather than logarithms of @xmath53 .
in many cases such aggressive growth would induce a failure of predictivity long before the end of inflation . also , whereas the coefficient which accompanies a logarithmic divergence is physically meaningful , and can be used as an input to renormalization group calculations , power - law divergences are by contrast mostly meaningless .
the late - time physics whose presence they signal must be found elsewhere : it can not usually be distilled from the properties of the divergences themselves .
although weinberg s theorem identifies a class of models containing interactions which may give rise to such fast divergences , it does not appear that any previously proposed inflationary model makes essential use of such interactions .
for this reason they have not yet received much attention . until non - linear questions became pressing ,
inflationary perturbation theory was dominated by the traditional lifshitz approach @xcite , in which one separates each field into a background @xmath54 and perturbation @xmath55 . at second order and above such calculations
become more arduous , although they have now been carried to an impressive degree of refinement @xcite .
the separate universe principle is an alternative to the lifshitz approach . according to this principle
, a volume of spacetime containing a background field @xmath54 and perturbation @xmath55 behaves ( on scales sufficiently large that gradients may be neglected ) just like an unperturbed universe containing the homogeneous field value @xmath56 .
if we solve for an arbitrary quantity @xmath57 with initial conditions set by the value of a background field at time @xmath31 in the unperturbed universe , we can determine the values taken by this quantity on superhorizon scales in the perturbed universe using the trivial identity @xmath58 where @xmath21 is the time at which we wish to evaluate @xmath59 .
this might typically be at the end of inflation or later .
whatever time we choose ,
. allows correlations of @xmath59 to be computed provided we know the correlation functions of @xmath60 at time @xmath31 . for example , the first contribution to the two - point function of @xmath59 is @xmath61 , where @xmath62 denotes the partial derivative of @xmath57 with respect to @xmath63 .
higher - order terms present in eq .
imply that other contributions must exist .
let us restore spatial arguments for clarity .
even if @xmath64 has gaussian statistics at time @xmath31 , there will be contributions of the form @xmath65 where @xmath66 denotes the preceding term with @xmath67 and @xmath68 exchanged and the symbol `` @xmath69 '' means that the correlation function contains this contribution among others .
indeed , these four - point correlations will be accompanied by six- , eight- and higher @xmath70-point correlation functions for all @xmath1 .
if @xmath64 has non - gaussian statistics then even more contributions will be generated .
zaballa , rodrguez & lyth introduced a set of diagrammatic ( `` feynman - like '' ) rules designed to keep track of these terms @xcite .
generalized rules were discussed by byrnes , koyama , sasaki & wands @xcite and in ref .
@xcite , but practical calculations are rarely of sufficient complexity to require them .
these contributions depend on the correlations among @xmath64 when @xmath71 becomes large , which are identified most simply in fourier space .
we find that @xmath72 receives contributions of the form @xmath73 in which @xmath74 is the power spectrum of @xmath64 , defined by @xmath75 in the limit @xmath76 , significant contributions to the @xmath77-integral in eq .
arise only from the region @xmath78 .
the behaviour of the integrand in the small-@xmath24 limit depends crucially on @xmath74 . for a scale - invariant power spectrum generated during inflation @xmath79 , which gives logarithmic singularities in the @xmath80-integral .
boubekeur & lyth @xcite , and later lyth & rodrguez @xcite , remarked that because the logarithmic behaviour is softened in the limit @xmath81 , the log - divergent part only receives contributions for @xmath82 .
therefore , to a reasonable approximation , @xmath83 where @xmath84 is an infrared cutoff .
this expression diverges in the limit @xmath85 .
it is clear at once that this divergence is not a physical prediction .
it depends on what is assumed about correlations among the perturbations in the infinite volume limit , where @xmath86 . even within the framework we are using , it is easy to see that if @xmath74 has constant red tilt then this logarithmic divergence is exchanged for a power law . if @xmath74 has constant blue tilt the integral is convergent .
an exactly logarithmic divergence occurs only for a scale invariant @xmath74 .
for example , if @xmath55 has even a very small mass this will give rise to a finite correlation length , beyond which correlations decay exponentially . the divergence in eq .
would then be absent .
we will see later that there are good physical reasons to believe this is what should happen in practice , at least in certain theories , and some evidence from concrete calculations that it does .
in any case we should also recognize that the framework we have been discussing is inadequate for the description of correlations on very large scales .
why is this ? to calculate the power spectrum from a model of inflation
we must typically assume that the background field @xmath54 is spatially homogeneous and depends only on time , whereas the spatially dependent fluctuation @xmath60 satisfies @xmath87 everywhere .
although this is reasonable on scales not too much larger than the de sitter horizon , it need not happen that every field admits such a decomposition as we pass to the infinite volume limit . if this is the case we should set the scale @xmath88 to be conservatively smaller than the largest scale for which a spatially homogeneous background @xmath54 can be found .
the calculation makes sense within this `` box . ''
if we are forced to discuss correlations on larger scales , they will have to be determined by patching together a mosaic of boxes in which the homogeneous background field may take different values .
this point of view was advocated by boubekeur & lyth @xcite , and later refined by byrnes , koyama , sasaki & wands @xcite , lyth @xcite and other authors @xcite .
this can not be the whole story , because eq . shows that the mosaicking procedure leaves behind terms involving @xmath89 .
these logarithms depend on the arbitrary scale @xmath88 . assuming the physics of the scalar field zero - mode to be quasi - classical on very large scales , the missing element is an accurate map of the average scalar field value within each box of the mosaic .
this map naturally depends on the scale @xmath88 , but the @xmath88 dependence of physical quantities cancels when we study correlations within the mosaic as a whole .
these issues will be discussed in more detail in [ sec : space ] .
we see again that the appearance of infrared divergences is connected with the existence of non - trivial background configurations . in the present case
the map could consist of an inventory of boxes , pairing each box with the average field value within .
alternatively , in a theory such as inflation where predictions are essentially statistical , it could simply consist of a probability distribution for the average field value , taken over the ensemble of boxes . in the previous two sections we have emphasized the role of the background field configuration in generating infrared divergences : when determining correlations among small fluctuations we take the background to be homogeneous and approximately time independent . in practice it may be neither , leading to the emergence of compensating logarithms .
there is another source of logarithmic corrections which is unconnected with the background field configuration .
suppose we define some quantum field theory at a scale @xmath90 . in using this field theory to compute correlations in vacuum where there is no background at all we are familiar with the appearance of `` large logarithms '' of the form @xmath91 , where @xmath92 is an energy scale characteristic of the correlation in question .
corrections of this sort occur in any quantum field theory , and the field theories we use to compute inflationary correlations are no exception .
such logarithms were studied by weinberg @xcite , who computed the correction induced by loops of @xmath26 different spectator fields in a model where inflation is driven by the vacuum energy associated with a single scalar field .
( 100,80 ) ( 100,80 ) [ see fig .
[ fig : oneloop ] .
in contrast to minkowski space , each external leg is evaluated at the _ same _ late time @xmath31 and therefore these diagrams should be read from the middle to the outside .
the interior of each diagram functions rather like an instanton . breaking the left - hand diagram down the middle ,
_ four _ quanta of the spectator field nucleate below the de sitter horizon , and propagate until one pair annihilates at time @xmath93 and a second pair at time @xmath94 .
the product of each annihilation is a quantum of the active field , which propagates freely until the surface @xmath95 ( represented by the outgoing arrows ) .
the shared history of these two particles generates a correlation . in an alternative and
equally acceptable interpretation , a pair of active quanta nucleate .
one particle propagates directly to the surface @xmath95 , whereas the second spontaneously fluctuates into a pair of spectator quanta at time @xmath93 , and at a later time @xmath94 coalesce to form another quantum of the active field before propagating to the final surface .
the right - hand diagram can be interpreted similarly , with the self - loop either providing a correction to the nucleation event , or dressing the propagation of a single quantum on its route to the surface @xmath95 . ] as in minkowski space , these logarithms arise from ultraviolet divergent momentum integrals which we can study by cutting off the integral for @xmath96 . for
a correlation dominated by wavenumbers of order @xmath24 , the typical integrals with which we are confronted take a form similar to @xmath97 where @xmath80 is the comoving three - momentum which circulates in the loop , and `` finite '' denotes terms which do not diverge in the limit @xmath98 . , rather than logarithms .
we are tacitly assuming that these divergences can be absorbed by renormalization - group running of the masses and couplings . ]
it is the appearance of @xmath77 in the combination @xmath99 which gives the logarithm a non - trivial @xmath24 dependence , and by the usual arguments we can be assured that the coefficient of the logarithm is independent of the precise ultraviolet regulator we choose .
this does not make eq . entirely unambiguous because its interpretation depends on the meaning we assign to @xmath100 , as we shall discuss shortly .
whatever its meaning , we see the emergence of the structure @xmath101 , with @xmath102 characterizing the scale of the correlation , and @xmath103 representing the point at which we define the theory .
these logarithms are a different species to the infrared logarithms of [ sec : zoo - time][sec : zoo - box ] , which diverged for fixed @xmath24 when the calculation was taken to occur in a box of infinite spatial or temporal extent . which accompanies the ultraviolet logarithm was erroneously identified with the scale @xmath88 which accompanies the box - cutoff logarithm .
this error was corrected from ` v3 ` onwards of the arxiv version of ref .
i would like to thank e. dimastrogiovanni for drawing this to my attention . ] eq .
is finite in this limit , although we will see below that quantum loops can also generate infrared divergent terms .
it is true that eq .
has bad behaviour when @xmath104 , but a divergence in this limit occurs even at tree - level . can be defined via normal ordering , @xmath105 where @xmath106 and @xmath107 may be singular as @xmath108 , and ` @xmath32 ' denotes terms which are finite in this limit . the difference between @xmath109 and the composite operator @xmath110 ( which has a finite vacuum expectation value ) accounts for the singularities of the wightman functions at coincidence . for an interacting field theory on curved spacetime
the normal - ordered product @xmath110 must be replaced by a more general definition of a renormalized composite operator , @xmath111 .
_ propose that such operators be defined by an adiabatic subtraction procedure rather than normal ordering .
( in principle , the curved space operator product expansion of hollands & wald @xcite could be applied instead . ) in their prescription , contributions to the classical power spectrum should be computed at horizon exit using the fourier modes of @xmath111 rather than @xmath109 , as is conventionally done . the adiabatic subtraction procedure which makes the renormalized composite operator @xmath112 well - defined in
the ultraviolet also entails subtraction of the low-@xmath24 modes which give infrared divergences , at least in the massless case @xcite . at the time of writing ,
most authors prefer to define the power spectrum using the fourier modes of @xmath109 .
] we nevertheless include these ultraviolet logarithms in this discussion of infrared effects because they alert us to the fact that large - scale ( that is , infrared ) structure in the background fields depends on physics at large values of the hubble rate , @xmath4 ( that is , in the ultraviolet ) .
this feature of gravity , which swaps ultraviolet and infrared physics , is known from previous applications of the holographic renormalization group to domain wall spacetimes which are asymptotically anti de sitter @xcite or de sitter @xcite .
we will return to this question when we revisit the mosaicking prescription outlined in [ sec : zoo - box ] , because it is clear that the large - scale structure of the mosaic will depend on what is assumed about the matter field theory in the ultraviolet .
we must still decide what meaning should be attached to @xmath100 .
in calculating loop corrections , weinberg @xcite made use of dimensional regularization .
this entails an analytic continuation to @xmath113 spatial dimensions , after which a limit must be extracted by dropping singular terms as @xmath114 @xcite .
later , chaicherdsakul @xcite and adshead , easther & lim @xcite employed the same technique . because @xmath100 does not appear explicitly in this method , being hidden in subtractions associated with the pole at @xmath115 , these papers quoted the loop correction as a multiple of @xmath116 but left the scale @xmath100 implicit .
. used an alternative procedure , imposing a sharp limit on the comoving momenta which contribute to each integral .
this method was employed in ref .
@xcite , and later by dimastrogiovanni & bartolo @xcite , and gives results in agreement with dimensional regularization . in either approach , @xmath100 must be interpreted as a comoving scale .
van der meulen & smit remarked that because comoving scales are not physical , logarithms of the form @xmath117 were not easy to interpret @xcite . it had previously been observed in ref .
@xcite that if one takes @xmath100 to be associated with a fixed physical scale @xmath118 , then at an arbitrary time the corresponding comoving scale is @xmath119 where @xmath120 is the scale factor .
but at which time should the cutoff be evaluated ?
senatore & zaldarriaga argued that an appropriate choice would be the time of horizon crossing @xcite , for which @xmath121 if @xmath122 is the value of the hubble parameter as the wavenumber @xmath24 crosses the horizon , at time @xmath123 . if so we could conclude that eq . ,
and other similar integrals , should be replaced by @xmath124 a similar conclusion had been reached earlier , by a different method , in ref . @xcite .
this would give a result in precise analogy with the logarithm @xmath91 encountered in scattering calculations , but seems incompatible with the use of dimensional regularization .
however , senatore & zaldarriaga went on to argue that the free - field propagators which appear in the feynman rules should also be calculated by analytic continuation of the mukhanov equation to @xmath125 spatial dimensions .
this was not attempted in refs .
analytically continuing in this way generates extra terms which enable dimensional regularization to reproduce . at the present time it seems unclear whether the horizon - crossing cutoff can be obtained from a more fundamental principle , or whether one must simply adopt it as a prescription giving reasonable results consistent with the approximate de sitter symmetry . in the comoving cutoff
was equivalent to including only quanta beneath the cutoff at the earliest of the times @xmath93 and @xmath94 in the left - hand diagram of fig .
[ fig : oneloop ] . in a more complicated diagram , one would choose the earliest such time .
essentially the same argument was given in ref .
@xcite , which discussed the influence of quanta which redshift under the horizon at times subsequent to @xmath93 . ]
it is also unclear how this form of dimensional regularization should be extended to momentum integrals arising from field redefinitions , which are to be discussed below . at one loop
these can be dealt with using a fixed momentum cutoff , so this ambiguity is not yet pressing . [ [ section ] ] for convenience , some results on loop corrections will be collected in this section .
weinberg gave his calculation in the comoving gauge , where the role of active scalar field was taken by the comoving curvature perturbation , @xmath126 .
adshead , easther & lim worked in the uniform curvature gauge , where the action describing a 3-point contact interaction between an active field @xmath54 and a collection of @xmath127 spectator fields @xmath128 can be written @xcite @xmath129 - \frac{1}{2 } \delta \phi s'_\alpha s'_\alpha - \partial^{-2 } \delta\phi ' s'_\alpha \partial^2 s_\alpha \right),\ ] ] where @xmath130 , summation of repeated @xmath131 indices is implied , a prime @xmath132 denotes a derivative with respect to conformal time and an overdot denotes a derivative with respect to cosmic time . in this equation ,
following the notation of maldacena @xcite , @xmath133 represents the first - order equation of motion for @xmath54 .
this term can be removed by a field redefinition is zero by construction ( at least to leading order ) when evaluated on a propagator and therefore gives no contribution whether we subtract it or not .
the @xmath133 term is a proxy for boundary terms , which following the notation of maldacena @xcite have not been written explicitly , which do _ not _ vanish on solutions to the equations of motion @xcite .
it was observed in ref .
@xcite that the boundary terms and @xmath133 terms are both removed by the field definition , allowing this notational trick to make sense . ]
@xmath134 the rules for evaluating correlation functions such as @xmath135 are discussed in the review paper by koyama elsewhere in this volume @xcite .
one finds the power spectrum of @xmath136 to be @xmath137 where now the time of observation , @xmath31 , is chosen to be almost immediately after the time of horizon crossing , @xmath138 . to obtain the two - point correlation function of @xmath60 , the effect of the field redefinition
must be reversed .
one finds @xmath139 where @xmath140 denotes the previous term with @xmath141 and @xmath142 exchanged , and @xmath143 is the convolution of @xmath144 and @xmath145 with argument @xmath77 . in evaluating this expression
we encounter three - point correlations of the form @xmath146 and four - point correlations of the form @xmath147 , but the four - point terms give no contribution to @xmath148 .
the three - point correlations can be evaluated using the general formula for a three - point function of scalar fields given in ref .
@xcite , yielding @xmath149 where @xmath150 . after some calculation
, one finds that the second term in brackets @xmath151 does not contribute , and the power spectrum of @xmath54 can be written @xmath152 from this calculation we see that field redefinitions reshuffle the coefficient of @xmath148 . since
gauge transformations are a form of field redefinition , we must expect the numerical coefficient @xmath153 in eq . to shift under a change of gauge .
in particular , this will happen if we change to comoving gauge and obtain the correlation function of the curvature perturbation , @xmath126 .
. can not be compared directly to weinberg s result in ref .
@xcite , but is an equally good indicator of the magnitude of the loop correction . some time after weinberg s @xmath126-gauge calculation of this spectator loop @xcite , the self - loop correction to the power spectrum of an active scalar field was obtained @xcite .
this corresponds to the right - hand diagram of fig .
[ fig : oneloop ] , and was found to give a correction to the power spectrum of the active field immediately after horizon crossing which amounted to @xmath154 when more than once active species is present one should also consider loops similar to the left - hand diagram of fig . [
fig : oneloop ] , with active fields circulating in the interior of the diagram . in principle
these allow the species of active scalars to fluctuate into each other , but have not yet been calculated .
[ [ section-1 ] ] dimastrogiovanni & bartolo obtained corrections from graviton loops @xcite , shown in fig .
[ fig : grav ] .
( 100,80 ) ( 100,80 ) before discussing these , consider the self - loop correction for an active scalar field .
wick contraction of an external field operator with an operator in the interior of the loop endows the right - hand diagram of fig .
[ fig : oneloop ] with a dependence on the external wavenumber .
this happens because operators like @xmath155 applied to the interior of the loop generate integrals similar to those in eq . , containing @xmath156 in the denominator .
in the right - hand diagram of fig . [
fig : grav ] the graviton pair interior to the loop must contract among themselves , and likewise the external fields .
therefore no dependence on the external wavenumber is generated , and this diagram factorizes into an overall renormalization of the tree - level two - point function , @xmath157 where @xmath88 is an infrared cutoff of the type which appears in box - cutoff logarithms , as discussed in [ sec : zoo - box ] . for the left - hand diagram , one finds .
dimastrogiovanni & bartolo found a slightly different coefficient for each logarithm .
as this paper was being completed , a recalculation of the graviton loop appeared in ref .
@xcite , which agrees with the result given here .
( p. adshead & e. dimastrogiovanni , personal communications . ) ] @xmath158 where `` @xmath32 '' denotes terms which vanish at late times together with any subleading contributions .
these examples are interesting because they show that quantum loops can generate `` box - cutoff '' effects in addition to `` new physics '' logarithms , which was suggested but not demonstrated in ref .
it is not yet clear how to interpret the association of a @xmath89 term with a loop containing spin-2 gravitational quanta , which lead to traceless deformations of the metric .
however , these terms all cancel in the total one - loop correction , as we now explain . together ,
eqs . , and
give the largest part of the loop correction for each active scalar field in a theory containing @xmath127 spectator fields .
the terms which are neglected by this combination are higher - order in @xmath159 but do not include compensating factors of @xmath127 .
first , the @xmath24-dependent logarithms in eqs . and cancel between themselves , to leave a logarithm depending only on the ultraviolet and infrared regulators .
second , this logarithm cancels with that in eq . ; only the contribution of eq . is left .
therefore , to this accuracy , the total loop correction is entirely free of the infrared regulator , @xmath88 , @xmath160 remarkably , this receives contributions from the isocurvature fields alone .
it is not clear whether this cancellation is accidental , or enforced by some deeper principle .
for example , in the case of a single minimally coupled scalar field in de sitter space ( obtained by passage to the limit @xmath114 ) , this absence of one - loop corrections becomes exact and is consistent with de sitter invariance .
if the cancellation is accidental there seems no reason why @xmath88-dependent logarithms should not appear at subleading order in @xmath42 , or at two loops and higher .
it would be of considerable interest to determine whether this is the case .
burgess _ et al . _
@xcite emphasized that , if infrared - sensitive contributions occur , one should obtain a time independent answer when working with _ physical _ ultraviolet and infrared cutoffs , @xmath161 and @xmath162 . in their analysis , this was achieved because the ultraviolet and infrared terms combined to give @xmath163 .
if one chooses the infrared cutoff to be comoving as we are doing in this discussion then de sitter invariance is broken and one can not avoid a time dependent logarithm of the form rejected by senatore & zaldarriaga as incompatible with eternal inflation . in the present framework ,
whether a similar problem exists depends on the time - dependent distribution of background field values which removes @xmath88 from eq .
and similar equations .
[ [ section-2 ] ] a somewhat less physical example is the loop in pure @xmath164 theory , which will play an interesting role in the discussion of time dependence in [ sec : time ] . taking the interaction to be @xmath165 one arrives at the loop of fig .
[ fig : cube ] .
( 100,80 ) van der meulen and smit calculated this loop , accounting for both ultraviolet and infrared effects @xcite .
it was later recalculated by various groups , since it provides a useful test case @xcite .
one finds @xmath166 [ [ section-3 ] ] a loop in @xmath167 theory was calculated by tsamis & woodard @xcite and petri @xcite , and later by burgess , holman , leblond & shandera @xcite .
the interaction is @xmath168 and the diagram is the same as the right - hand loop of fig .
[ fig : oneloop ] , although with a different vertex .
the result is @xmath169 other examples of loops have been considered in the literature . in ref .
@xcite a set of loop corrections to the two- and three - point functions of an active scalar field were calculated in a theory where the scalars coupled to the field strength of an abelian gauge field . although their structure is rather complicated , these loops gave results which are not qualitatively different to those we have already studied .
as in any quantum field theory , the diagrams in figs . [
fig : oneloop][fig : cube ] are calculated by identifying quasi - free propagating modes which travel along the arcs of the diagram and interact at vertices . in each calculation
discussed above , the quasi - free modes behave like massless scalar fields in eternal de sitter space which have two - point correlations of the form @xmath170 where @xmath171 and @xmath172 .
the magnitudes are equal on - shell , where conservation of momentum is satisfied , so @xmath173 . the distinction has been maintained in eq .
because in practical calculations one must often calculate off - shell where @xmath174 , only taking the on - shell limit at the end of the calculation @xcite ; the same is true for higher @xmath1-point functions where one must keep the @xmath175 all distinct .
eq . is built out of elementary wavefunctions proportional to @xmath176 , which asymptotes to unity in the infinite future of de sitter space where @xmath177 .
therefore , once correlations have been generated using eq .
they persist into the indefinite future .
had we instead taken the quasi - free modes to have a mass of order @xmath178 , we would have obtained elementary wavefunctions of the form @xmath179 , where @xmath180 are hankel functions of the first and second kind , respectively , and the order of these functions , @xmath181 , is related to the mass @xmath178 by @xmath182 for @xmath183 this gives the hankel functions an imaginary order . in the limit @xmath184 the massive de sitter wavefunction would have an asymptotic expansion proportional to @xmath185 where `` @xmath32 '' denotes terms which converge to zero more quickly than those which have been written .
clearly , correlations are suppressed at late times whenever @xmath186 and vanish entirely as @xmath177 .
having obtained expressions of this form , riotto & sloth @xcite observed that eq . expresses nothing more profound than that a field of mass @xmath178 has a proper correlation length of order @xmath187 , and in the infinite future of de sitter space the cosmological expansion has carried any two spatially distinct points far beyond a proper separation of order @xmath187 .
for this reason , correlations between spatially separated points must decay .
weinberg remarked that this mismatch between eq .
and the decay of correlations in the real universe compels us to account for mass terms non - perturbatively @xcite . in principle
we could deal with a term in the lagrangian of the form @xmath188 by including its contribution in the elementary wavefunctions of the quasi - free modes , or by counting it among the interactions . if we include it among the interactions , then we compute correlations using eq . and
are obliged to make the implicit assumption that they do not subsequently decay .
this may yield reasonable results , supposing the correlations evolve only slowly , provided we do not ask for their properties too long after horizon crossing .
nevertheless , where the correlations truly decay this approximation must lead to trouble at late times .
since we expect most fields to acquire a non - trivial potential except where protected by an exact symmetry , we also expect the fluctuations in these fields to develop small masses . as was explained in
[ sec : zoo - time ] , it is our failure to account for this time evolution in the background fields , and the subsequent decay of correlations , which is responsible for the appearance of the secular logarithms @xmath36 .
what if a field has a potential containing no mass term , but other higher - order interactions ?
this situation was analysed by senatore & zaldarriaga @xcite , who argued that on any potential of this kind the background field would still roll down and evolve .
for this reason , we would expect a mass term to be generated for the fluctuations after resumming a sufficient number of insertions in the propagator .
this would subsequently cause correlations to decay .
therefore this situation is qualitatively the same as the case of a mass term in the potential .
we will see an explicit example in [ sec : dynamical ] .
how fast can we expect correlations to evolve ? in [ sec : zoo - time ] we discussed a theorem due to weinberg @xcite which guarantees that only logarithms of @xmath53 appear in inflationary correlation functions , and not power laws of the form @xmath189 for @xmath190 which in principle are allowed @xcite .
( contributions proportional to positive powers of @xmath53 are generically present , but contribute nothing at late times . )
the precise statement of this theorem applies to scalar and tensor excitations . in three dimensions
, we have : in three space dimensions , power law divergences are absent at late times in correlations among scalar and tensor quanta , provided all interactions in the lagrangian , when written in cosmic time , fall into one of two classes : 1 .
interactions containing strictly less than one factor of @xmath52 .
interactions which may grow as fast as @xmath52 ( but no faster ) , and which contain only fields rather than time derivatives of fields .
weinberg proved this theorem by studying commutators among the fields and their derivatives at late times , and then counting how these commutators could appear in the possible elementary interactions .
this method of proof does not make use of the explicit slow - roll , massless solution .
however , senatore & zaldarriaga later remarked that a refinement of this theorem might be possible @xcite , because the asymptotic estimates obtained by weinberg _ did _ assume that the fields were massless .
in addition , since each elementary interaction was considered in isolation , the possibility of cancellations among groups of interactions was not considered . at the time of writing , these possible refinements have not been studied . in a later publication ,
weinberg extended this theorem to abelian and non - abelian vector fields , and dirac fermions @xcite ( see also ref .
these fields have rather better behaviour than the scalar fields considered in the original theorem , and correlation functions among them typically converge at late times . for the remainder of this discussion
, we will restrict attention to theories satisfying the conditions of weinberg s theorem .
therefore only secular logarithms will appear as @xmath37 , and not the faster power - law divergences .
up to this point we have not made our choice of gauge explicit , but we have been discussing fluctuations in scalar fields .
this will usually correspond to the uniform curvature gauge .
the quantity whose correlations are accessible to observation in the cmb is often taken to be the comoving curvature perturbation , @xmath126 .
( indeed , weinberg s theorem and the spectator loop discussed in [ sec : zoo - new ] were given directly in this gauge . ) the statistical properties of @xmath126 are determined from a superposition of the active fields in the theory .
many prescriptions have been employed to obtain this superposition @xcite , but a simple method is to observe that @xmath126 is a local fluctuation in the aggregate cosmological expansion , @xmath192 , where @xmath120 is the scale factor .
we can therefore use the separate universe formula , eq . .
one usually defines @xmath193 , where @xmath26 is the number of e - folds from a reference time where @xmath194 .
we take the initial slice to be a uniform curvature hypersurface , and the final slice to be a surface of uniform energy density , after which we arrive at the @xmath191 formula @xcite , reviewed elsewhere in this volume by tanaka , suyama & yokoyama @xcite and wands @xcite @xmath195 among other things , this is a gauge transformation .
the @xmath191 formula reproduces those terms from the non - linear transformation between the uniform curvature gauge and comoving gauge @xcite which do not vanish on superhorizon scales .
. enables the correlation functions of @xmath126 to be determined from those of @xmath64 , which in turn are computed using the massless propagator . in the foregoing discussion we argued that eq .
prevents correlations from decaying after they have been generated , and that this corresponds to neglecting evolution of the background fields .
the @xmath191 formula ameliorates this difficulty .
suppose we wish to compute correlations on some scale @xmath24 .
if we take the initial time , @xmath31 , to be shortly after the time of horizon exit associated with @xmath24 , then we do not commit a gross error in failing to account for their subsequent decay . moreover , the time dependence summarized by @xmath26 accounts for evolution of the background fields as they roll down their potentials .
therefore we expect eq . to incorporate the infrared physics whose neglect gave rise to secular logarithms .
this argument was given in ref .
@xcite and is supported by the analysis of van der meulen & smit @xcite , although a formal proof is still lacking .
the same conclusion can be reached by considering the logarithms themselves .
if @xmath31 is chosen not long after horizon crossing , then the terms @xmath36 are of order unity .
if the leading term containing @xmath1 powers of @xmath36 occurs in the combination @xmath196 , then , not long after horizon crossing , all terms containing logarithms will be small compared to the tree level whenever @xmath197 .
therefore we can neglect all logarithms in using eq .
to compute correlations of @xmath126 .
the effect has been to `` resum '' their effect into the background evolution of the scalar fields , which is described by the accumulating number of e - folds , @xmath26 .
for example , in a single - field model of inflation this argument reproduces the expected conclusion that @xmath126 does not evolve .
but the virtue of the @xmath191 method is that it is not restricted to the single field case : in a multiple field scenario where @xmath126 has non - trivial evolution , the @xmath191 formula will correctly describe the growth and decay of fluctuations
. there may be problems when using eq . to obtain the correlation functions of @xmath126 if one wishes to include the `` quantum '' logarithms of [ sec : zoo - new ] .
this procedure was applied in refs .
@xcite . however , as we have already observed , the @xmath191 formula is a gauge transformation , and it was argued in [ sec : zoo - new ] that such transformations will typically modify the coefficient of @xmath198 .
although this possibility has yet to be investigated in detail , the large number of ultraviolet terms discarded by the @xmath191 formula ( which should be compared , for example , with eq .
( a.8 ) of ref .
@xcite ) would apparently make a mismatch likely .
the @xmath191 formula retains all relevant infrared physics , so we would not expect similar difficulties with secular and box - cutoff logarithms . for the purpose of comparison with observation , the @xmath191 method is presumably sufficient and allows us to extract initial conditions from typical inflationary models without any loss of predictivity . in many models of inflation
we wish to obtain the properties of correlations only 50 to 60 e - folds after they are synthesized . in a conventional model
the only source of appreciable time evolution during these 50 to 60 e - folds comes from the classical variation of the background . nevertheless , as a point of principle , one might prefer a more satisfactory resolution .
some time after the resummation interpretation of @xmath191 had been proposed , burgess , holman , leblond & shandera remarked that it did not account for the possibility of secular time evolution from quantum effects @xcite , such as those of eqs . , and .
these terms typically appear in combination with the small parameter @xmath3 , and are therefore small unless many e - folds have elapsed since horizon crossing .
burgess _ et al .
_ suggested that these logarithms could be resummed using the method of the renormalization group , and applied this technique to the quartic loop of eq . .
we briefly review their method .
the starting point is to introduce an arbitrary constant , @xmath12 , which is unity in the tree - level power spectrum , @xmath199 our strategy is to resum the large logarithms into @xmath12 .
to do so , rewrite eq . in terms of an arbitrary intermediate time scale @xmath200 , @xmath201 \right ) .\ ]
] clearly , the combination of overall prefactor and first bracket is of the same form as @xmath202 , with @xmath203 .
we can therefore replace one by the other .
whatever the result , it must be independent of the arbitrary time @xmath200 . applying the familiar method of gell - mann and low @xcite
, we find @xmath204 solving this differential equation and setting @xmath205 , we conclude @xmath206 \right ) , \label{eq : burgess - resum}\ ] ] where @xmath207 satisfies @xmath208 the error in this procedure is of order @xmath209
. therefore eqs
. are trustworthy even when @xmath210 , giving rise to a resummation of the leading logarithms .
subleading logarithms could be included by accounting for terms of order @xmath211 or higher in eq . .
. describes decaying correlations . on the basis of the foregoing discussion
, we expect this behaviour to represent the correct infrared physics in a large class of theories . indeed ,
comparison with eq .
shows that the effect of @xmath212 is to introduce a dependence on @xmath213 comparable to the case of a massive scalar field . making use of eq . and
recalling that the square of eq .
yields the power spectrum , we see that eq .
is equivalent to a dynamically generated mass @xmath214 @xcite , @xmath215 if @xmath216 and @xmath88 do not generate an exponential hierarchy and @xmath217 , then @xmath214 is of order the hubble rate .
we can regard this as a concrete example of the argument of senatore & zaldarriaga @xcite discussed just before [ sec : weinberg ] .
this conclusion does not apply to every theory .
burgess , holman , leblond & shandera observed that the same argument applied to pure @xmath164 theory does not yield a dynamically generated mass .
the obstruction comes from the power of @xmath36 in the leading term of eq . .
ignoring the ultraviolet logarithm proportional to @xmath198 , which plays no role here , resumming the secular logarithms would give @xmath218 \right ) .
\label{eq : cubedrg}\ ] ] since this does not suppress the power spectrum by a positive power of @xmath213 at late times we can not interpret it in terms of an effective value of @xmath181 in eq . , and thus an effective mass @xmath214 .
indeed , has rather poor behaviour in the limit @xmath219 , because the exponential diverges there .
one should not take this behaviour literally .
included logarithms of cubic order which have been neglected here , and which would become negative in the limit @xmath220 .
such terms would compete with the positive contribtion from @xmath221 .
thus , in pure @xmath164 theory the behaviour of the exponential in the far future becomes a delicate question .
this is not difficult to interpret . in pure @xmath164 theory
the hamiltonian is unbounded below , and there is nothing to prevent fluctuations ( and their correlations ) growing indefinitely .
_ remarked that this pathological behaviour made studying the late - time behaviour of this theory problematic . following the argument of senatore & zaldarriaga @xcite
, we may expect that in realistic theories behaviour comparable to is more generic .
less is known about the resolution of what we have called box - cutoff logarithms , of the form @xmath89 , which were left behind in [ sec : zoo - box ] after tiling the spatial region of interest into a mosaic of boxes . our discussion in [ sec : zoo - box ]
was based on an assumption : that the role of physics at scales larger than @xmath88 was to determine a quasi - classical background field configuration in each box .
a proof of the validity of this assumption is not known .
what would it entail ? consider an expectation value of local operators @xmath222 , , @xmath223 evaluated at spatial positions @xmath67 , , @xmath224 with roughly common separations @xmath225 which are small on the characteristic scale of the mosaic , so that @xmath226 .
we wish to evaluate this correlation function at time @xmath31 , just after horizon exit of the wavenumber @xmath24 , which can be achieved by averaging against some wavefunctional @xmath227 $ ] weighting spatial configurations of the @xmath228 @xcite , @xmath229 \ ; o_1({\bm{\mathrm{{x}}}}_1 ) \cdots o_n({\bm{\mathrm{{x}}}}_n ) \ ; | \psi_\ast [ o ] |^2 , \label{eq : expectation - value}\ ] ] where a subscript ` @xmath30 ' denotes evaluation at @xmath31 .
. leads to the in in or schwinger path integral once each wavefunctional has been written as an integral over all histories @xmath230 with appropriate boundary conditions at @xmath31 .
cut out a box of comoving size @xmath231 , approximately centred on the @xmath232 .
the path integral divides into separate integrals over the spatial field configuration interior to the box , written @xmath233 , and the exterior configuration , @xmath234 .
these configurations merge continuously on surfaces of the box , denoted @xmath235 . in these terms , eq .
can be rewritten @xmath236 \ ;
o^-_1({\bm{\mathrm{{x}}}}_1 ) \cdots
o^-_n({\bm{\mathrm{{x}}}}_n ) \ ; | \psi^-_\ast [ o^- , o^-_{\partial l } ] |^2 .\ ] ] the interior wavefunctional @xmath237 satisfies @xmath238 = \int [ { \mathrm{d}}o^-(\tau ) ] \ ; \exp \left ( { \mathrm{i}}s[o^- ] \right ) \int [ { \mathrm{d}}o^+(\tau ) ] \ ; \exp \left ( { \mathrm{i}}s [ o^+ ] \right ) , \label{eq : factorized - path - integral}\ ] ] where @xmath239 is a local action functional . in this expression , @xmath240 $ ] represents an integral over interior field histories @xmath241 which coincide with @xmath242 at @xmath31 and have boundary data @xmath243 on @xmath235 , whereas @xmath244 $ ] represents an otherwise unconstrained integral over all exterior field histories @xmath245 which merge sufficiently smoothly with the boundary configuration @xmath243 on the edges of the box . in typical applications ,
the expectation value does not depend on the prior history of the universe up to the point of horizon crossing .
it is therefore independent of almost all boundary data represented by @xmath243 , retaining only a possible dependence on the boundary configuration near the surface @xmath95 .
if the path integral is dominated by semiclassical field configurations which vary only slowly on the scale @xmath88 , then to a good approximation this remaining dependence will involve only the roughly homogeneous boundary value at @xmath31 , and not its spatial gradients .
. suggests that expectation values are to be calculated by fixing a homogeneous background field configuration at the surface of horizon crossing , and summing over unrestricted field histories for fluctuations in the interior .
finally , one averages over the fixed background configurations with an unknown probability measure , @xmath246 , which derives from the integral over exterior field configurations . at this point
we have arrived at the prescription of [ sec : zoo - box ] .
this construction makes @xmath246 a distribution over field configurations at a fixed spatial position in an ensemble of realizations , but if ergodicity applies we can equally interpret it as a probability distribution over different boxes in the same realization .
the above argument is merely heuristic , and the existence of such a decoupling limit has not yet been demonstrated .
nevertheless , eqs .
show a strong similarity with path integral approaches to the operator product expansion ( `` ope '' ) in flat space quantum field theory @xcite .
a manifestly local and covariant formulation of this expansion in curved spacetime has been given by hollands & wald @xcite , which in principle could be applied to the study of infrared effects in nearly de sitter spacetimes .
however , this has not yet been done .
a similar conclusion was reached by allen & folacci @xcite and later kumar , leblond & rajaraman @xcite , who studied ambiguities in defining the zero - mode for the bunch
davies propagator associated with a massless , minimally coupled scalar field in de sitter space .
_ noted that similar ambiguities were removed in minkowski space owing to requirements imposed by the principle of cluster decomposition . exploiting the extra freedom available in de sitter space , they succeeded in subtracting infrared divergences by performing an appropriate redefinition . in a field theory
this redefinition would naturally become spatially dependent on scales which are large compared with those appearing in expectation values .
the result is to reproduce the conclusion that infrared divergences derive from a supposition that calculations are to be carried out with a definite , homogeneous expectation value for the background field configuration . from this point of view , eq . can be thought of as an analogue of cluster decomposition in de sitter .
these ideas have a large literature of their own .
kirsten & garriga @xcite used related reasoning to represent the quantized massless scalar field as the product of a hilbert space , associated with the zero - mode , and a fock space corresponding to excitations of finite wavenumber .
related ideas were explored by moncrief @xcite , higuchi and collaborators @xcite , and later giddings & marolf @xcite .
urakawa & maeda investigated the same physics from a different perspective @xcite . while discussing logarithms generated by ultraviolet effects , senatore & zaldarriaga remarked that combinations such as @xmath117 in eq .
were unphysical @xcite , because there is no physical comoving scale which could accompany @xmath24 .
however , this argument does not entirely preclude the appearance of @xmath24 in any logarithm .
for example , the combination @xmath36 measures by how many e - folds the wavenumber @xmath24 is outside the horizon at time @xmath31 and is a sensible physical quantity .
nevertheless , this example shows that one must be careful in dealing with the combination @xmath89 , in which @xmath88 is also a comoving scale .
let us first idealize to an inflating volume which is spatially infinite .
as in [ sec : zoo - box ] we divide this volume into boxes of size @xmath88 , in each of which the field is spatially homogeneous with small perturbations .
we have seen that correlations computed within these boxes depend on the arbitrary scale @xmath88 , but ( assuming only zero - modes need be retained from the long wavelength physics ) an answer independent of @xmath88 is obtained if we average the correlations over boxes @xcite .
this requires a knowledge of the homogeneous background field within each box of the mosaic .
if merely probabilistic answers are acceptable , it may only be necessary to use information about the statistical distribution of field values over the ensemble of boxes .
cosmological perturbation theory has been discussed from this point of view by many authors ; see , for example , the recent textbook by mukhanov @xcite .
if we assume that only zero - modes need be retained , as in
[ sec : mosaic ] , it follows that this argument does not depend on a careful discussion of how the total @xmath88 dependence can somehow cancel between the @xmath88-dependent box - sized correlations and the distribution of field values among boxes . the schematic argument of eqs .
shows that in a full treatment , where the entire ultraviolet physics and initial conditions are known , we have done nothing more than divide the calculation into two parts .
when we assemble these two halves , the final answer must be independent of @xmath88 .
this happens automatically , no matter how remarkably fine - tuned the necessary cancellations appear , and is a standard argument in the application of effective field theories @xcite .
the scale @xmath88 therefore disappears from the answer and we need not worry about its physical significance . in practice , the distribution of field values among boxes of the mosaic will evolve in time .
after convolving with this distribution , @xmath88-dependent terms would be traded for a time dependence which could itself be represented by secular logarithms
. however , it may not be the case that this @xmath88-independent answer is the quantity we wish to compare with observation . in discussing similar questions almost twenty years ago
, salopek & bond remarked that we only wish to calculate what we can observe @xcite .
what can we observe ?
only the density fluctuation within our presently observable patch of the universe @xcite .
observationally it is a matter of perfect indifference to us ( although , of course , of surpassing theoretical interest ) whether the distant , unobserved universe contains many regions similar to our own , or whether all distant regions are quite dissimilar .
whichever is correct , the answer will not change the result of satellite observations made today . if we subscribe to programme outlined in
[ sec : intro ] , then we must suppose that our observable region of the universe passed through a phase of inflation which eventually came to an end .
we assume that the classical background associated with whatever scalar fields supplied the necessary vacuum energy became roughly homogeneous within a box somewhat larger than our present horizon .
this box should be large enough to contain many regions of comparable size to the presently observable universe , but need not be exponentially larger @xcite .
we wish to compute correlations in the density fluctuation which would be recorded by a typical observer whose local patch experienced the same gross cosmological history . for this reason it may be _ incorrect _ to compare the @xmath88-independent answer discussed above with experiment , because in this answer
the correlations specific to our patch are commingled with correlations recorded by observers experiencing a gross cosmological history quite different to our own .
therefore , let us choose @xmath88 so that @xmath247 , making @xmath88 a little larger than our presently observable universe .
was suggested by boubekeur & lyth @xcite , who referred to it as a ` minimal box . '
its properties were later studied in a series of papers by lyth and collaborators @xcite . the same prescription , in various forms , was given by bartolo _
_ @xcite and later by enqvist _
_ @xcite and kumar _
_ @xcite . ]
the problem we face in this scenario is to determine the correct homogeneous background in which we should carry out our calculation . in general the answer depends on the distribution of field values within the ensemble of boxes , but there is a simple class of models in which this question is trivial . in any single - field model with a unique reheating minimum there is a unique value of the background field , @xmath248 , when inflation is of order 60 e - folds from ending .
the prior history of the large - scale universe is irrelevant @xcite . to compute what would be recorded by a typical observer in this model , we are entitled to carry out our calculation within a box carrying an approximately homogeneous background field with value @xmath249 . in this picture , the effective field theories we use to study slow - roll inflation are to be understood only as a description of the process by which an inflationary trajectory arrives in its final reheating minimum , causing inflation to end
. in the distant future our observable region of the universe will expand to include modes which are presently unobservable .
enqvist , nurmi , podolsky & rigopoulos remarked that an inflationary theorist , calculating millions of years in the future , would be obliged to work in a box of slightly larger size @xmath250 @xcite .
the argument of boubekeur & lyth @xcite , which was extended to two loops by bartolo _
@xcite , shows that physical observables compose correctly under the process of averaging @xmath88-sized boxes to make an @xmath251-sized box . in a scenario which is more general than single - field inflation there may be many terminal vacua in which the universe can reheat , rather than a unique choice . in these models
there is no preferred background configuration analogous to the single - field configuration @xmath249 in which we should carry out our computations .
even where a terminal vacuum can be selected in advance , it was shown by byrnes , choi & hall that the final fluctuations may depend sensitively on the path by which the fields arrive at the minimum @xcite ( reviewed in ref .
@xcite ) . to determine what would be recorded by a typical observer we must know at least the frequency with which these minima are populated by boxes in which inflation terminates .
this can be ascertained only if we know the distribution of field values within boxes of the mosaic , and ( as we have discussed above ) this distribution is an ultraviolet - dependent quantity . without knowledge of the complete ultraviolet physics , including any relevant initial conditions
, we can not calculate it from first principles .
apart from our interest in observational predictions , there may be other reasons to enquire about the large - scale disposition of the scalar fields .
senatore & zaldarriaga @xcite emphasized that we would like to know the circumstances under which eternal inflation can occur @xcite .
it is possible this depends on effects arising from loops at high order in perturbation theory .
for this reason we may wish to go beyond the narrow view advanced above , according to which our theories of slow - roll inflation should be restricted to a description of how inflation ends .
using a model of slow - roll inflation to describe evolution within a large spatial volume over very long times would again require us to know the distribution of field values over the entire mosaic of boxes .
clearly we can not hope to calculate this distribution _
ab initio_. suppose , however , that by some means we know the distribution of field values over the mosaic of boxes at time @xmath252 , at which point we suppose the hubble scale took the value @xmath253 .
if we have an effective description of the processes by which correlations are established at energies lower than @xmath253 , then we _ can _ calculate the subsequent evolution of the mosaic .
similar techniques are used in many applications where the long - range physics likewise can not be determined from first principles .
a simple example is the distribution of partons within colliding nuclei @xcite .
since this distribution depends on the unknown details of confinement in qcd it can not be calculated from first principles .
however , its evolution with energy can be calculated at energies large enough that the perturbative regime of qcd is a good description . in the inflationary case
, it is the regime of small @xmath254 which is accessible to perturbative calculations .
a simple method of calculating this evolution was suggested by starobinsky @xcite and later applied to the calculation of correlation functions in the deep infrared by starobinsky & yokoyama @xcite . in starobinsky
s method we suppose that the background field value within a given box of the mosaic is known .
our discussion of time evolution in [ sec : zoo - time ] and [ sec : time ] shows that we will obtain reasonable answers only if we account for the slow roll - down of this field . in a single - field scenario
the evolution of the background field value is given by @xmath255 where @xmath256 is the potential and @xmath257 is a gaussian noise term satisfying @xmath258 in a careful treatment , we would find that this noise term depends on what we assume about the physics responsible for generating correlations within this box .
. is a langevin equation , which can be equivalently expressed as an evoution equation for the distribution function of field values over the mosaic , @xmath259 , @xmath260 various refinements of this equation are possible , which were discussed by salopek & bond @xcite .
it has been suggested that stochastic inflation may represent one possible method of determining the distribution of field values over the entire mosaic of boxes @xcite .
although this idea is attractive , there are several drawbacks .
first , eqs .
depend on an initial condition for the distribution @xmath259 .
this initial condition is sensitive to the details of ultraviolet physics and is no more calculable from first principles than is @xmath259 itself .
this difficulty is usually circumvented by seeking stationary solutions of the fokker
planck equation , eq . .
however , it is not clear under which circumstances this is a reasonable choice , or whether the result genuinely approximates what would be obtained if we were to retain the complete ultraviolet physics .
second , the evolution equations and depend on what we assume about physics at high values of the hubble parameter , up to @xmath261 . in particular , these equations assume that slow - roll inflation is a good description throughout this regime .
this is analogous to the supposition that the standard model is a good description of physics at energies from the electroweak frontier at @xmath262 to up the gut scale at @xmath263 .
this assumption allows renormalization group evolution of the su(2 ) , u(1)@xmath264 and strong coupling constants to predict a unification near the gut scale in the same way that eq .
allows us to predict the behaviour of @xmath54 at values of @xmath4 far above those which gave rise to observable density fluctuations . in either case
it is perfectly reasonable that unguessed physics intervenes to spoil the conclusion .
if so , the true evolution of the background fields might follow laws very different to eqs . and .
alternatively , at sufficiently high values of @xmath4 the concept of a homogeneous background field itself may lose meaning . in that case the present discussion would be invalidated , becoming relevant only at lower energies .
it was emphasized above that , assuming only zero modes need be retained from the quasi - classical physics on large scales , cancellation of @xmath88 was automatic after averaging over the distribution of field values within the mosaic .
since there is no recipe to construct this distribution , it must checked individually for each candidate .
unfortunately , it is non - trivial to verify that the @xmath88-dependence carried by a solution to eq . would precisely cancel that @xmath88-dependence of correlations computed within a box @xmath88 . at the time of writing
, this does not appear to have been verified .
a complete cancellation would depend on retaining sufficient information about the relevant long - range physics in eqs .
, and it is far from obvious that this is the case .
however , since the same physics which gives rise to the box - cutoff logarithms is used to compute the evolution of @xmath54 in eq . , it is at least plausible that the @xmath88 dependence of @xmath265 may have the requisite properties .
a similar property is exhibited by the parton distribution functions which describe the physics internal to colliding nuclei @xcite . a first step in this direction
was given by kuhnel & schwarz @xcite who argued that large - scale correlations were strongly suppressed by stochastic effects , leading to infrared finite correlation functions .
in [ sec : zoo - time ] and [ sec : time ] we discussed the role of an evolving background field in suppressing correlations at late times , and in [ sec : dynamical ] we reviewed the argument of burgess _
@xcite , according to which the secular logarithms could be resummed into a dynamically generated mass .
this mass drives any correlations to zero in the far future , as the field settles in its minimum and loses memory of its past history .
eq . and
its associated fokker
planck equation , eq .
, account for both fluctuations and time evolution in the background field .
a fluctuation @xmath266 in eq . can push the field up the potential . at later times
the term @xmath267 causes it to roll towards the minimum .
therefore we would not expect statistical properties computed using the stochastic framework to suffer from the pathological failure to decay associated with correlations established using eq . .
this expectation is borne out in explicit calculations .
starobinsky & yokoyama worked with a model of @xmath268 inflation , defined by the interaction of eq . , and calculated the evolution of correlations in this theory using the stochastic method @xcite . working in the massless case for simplicity
, they were able to use the fokker
planck equation to derive an evolution equation for the two - point function @xmath269 , @xmath270 starobinsky & yokoyama observed that the solution of this equation gave a constant , @xmath271 at late times , and remarked that this was equivalent to the dynamical generation of a mass of order @xmath4 .
we briefly review this conclusion using the analysis of riotto & sloth @xcite , who derived a `` gap '' equation for the @xmath54 propagator , @xmath107 , @xmath272 \right ) g(x , x ' ) = \frac{{\mathrm{i}}}{\sqrt{-g } } \delta ( x - x ' ) \label{eq : gap}\ ] ] where the differential operator @xmath273 operates on the variable @xmath274 , and @xmath275 is a covariant derivative compatible with the metric @xmath276 .
eq . is obtained using the method of the 2-particle irreducible action . for details
the reader may consult the original article by riotto & sloth @xcite , or the textbook by calzetta & hu @xcite .
after translating to fourier space and substituting starobinsky & yokoyama s late - time solution for @xmath269 , given by eq .
, in eq .
one finds that the the green s function @xmath277 should be built out of solutions to the homogeneous equation @xmath278 this is the klein
gordon equation for a field of mass @xmath279 later , burgess _
_ @xcite argued that this mass could be reproduced from the dynamical mass generated by the renormalization group , eq .
, at least in the large-@xmath26 limit of an @xmath280-symmetric theory similar to that studied by riotto & sloth @xcite and petri @xcite .
if an exact shift symmetry protects the potential of some field , then it experiences no classical evolution .
an example might be a goldstone boson or axion , or more generally any isocurvature field which can be interpreted as an angle .
once perturbations have been generated in such a field , they do not decay . in this case
we would not expect resummation to generate a mass , and there is no prediction for the value taken by such a field within our presently observable patch , although it may be possible to estimate a distribution @xcite .
hertzberg , tegmark & wilczek @xcite argued that such scenarios represented a test case of anthropic reasoning . what would happen if a microwave background observable depended on the value of such a field ?
we would not be able to predict which measurements should be recorded by cmb survey satellites , such as the wmap and planck satellite missions . a similar argument was given by sloth , who pointed out that an observable such as the tensor fraction , @xmath281 , may depend on a ratio of two quantities which receive loop corrections scaling differently as the box size increases @xcite .
the expected value of @xmath281 will depend on the box in which correlations are computed .
however , as has been stressed throughout , to obtain observables one must compute a conditional expectation value , contingent on inflation ending in our own vacuum ( almost surely ) everywhere within the box .
for this reason , the observational relevance of infrared corrections in @xmath281 is not yet clear .
infrared effects have attracted considerable interest , but at the time of writing it has not been possible to extract a definitive prediction of a novel effect from these calculations that is , an effect which is not already implicit in calculations carried out locally around background fields set at a point on the scalar potential .
the best studied examples of infrared effects are the time - dependent secular logarithms introduced in [ sec : zoo - time ] . over short timescales
these logarithms represent evolution of correlations , as the background fields slowly roll down their potentials . at later times quantum effects can become important , but their role remains somewhat unclear .
the calculation of burgess _
et al . _ , supported by the arguments of senatore & zaldarriaga , suggests that in stable theories they will suppress correlations at late times .
box - cutoff logarithms represent a different problem : the possibility that , after many e - folds of inflation , spatial gradients develop in the background scalar fields .
if we only wish to use our theories of inflation to study observable correlations generated on approach to reheating , then this may not be a significant problem .
on the other hand , if we wish to study the evolution of a large spatial volume undergoing inflation for a long period of time , we expose ourselves to the possibility of unknown new physics at large values of @xmath254 .
the gauge invariant arguments of losic & unruh appear to cast doubt on the idea that perturbation theory remains globally well defined for an indefinite period in de sitter or nearly de sitter spacetimes . except in special cases , such as single - field inflation
, the study of correlations generated near exit from inflation is still ultraviolet - sensitive ( in the sense of physics at large values of @xmath254 ) and infrared - sensitive ( in the sense of physics on very large lengthscales ) , because we do not know in which reheating minimum inflation is likely to terminate .
this need not be a problem if observation can accurately determine the properties of our local minimum . in that case
we would be able to understand all the properties of our vacuum , but we would not be able to understand how to embed this minimum in the larger landscape of the inflationary potential .
however , this problem is not new and has been with us since the early days of the inflationary paradigm .
my understanding of the ideas discussed in this article is due to discussions with my colleagues and collaborators .
i would like to thank peter adshead , nicola bartolo , christian byrnes , xingang chen , emanuela dimastrogiovanni , richard easther , louis leblond , james lidsey , eugene lim , bojan losic , david lyth , karim malik , sami nurmi , dmitry podolsky , antonio riotto , sarah shandera , martin sloth , andrew tolley , filippo vernizzi and daniel wesley for sharing many insights concerning inflation generally , and the treatment of de sitter loop diagrams in particular .
clifford burgess , xingang chen , richard holman , louis leblond , sarah shandera and martin sloth offered helpful criticisms on an early typescript of this review .
i would like to thank peter adshead for a significant investment of time in verifying my recomputations of the inflaton self - loop , spectator - loop and graviton - loop diagrams discussed in [ sec : zoo - new ] .
100 t. falk , r. rangarajan , and m. srednicki , _ the angular dependence of the three point correlation function of the cosmic microwave background radiation as predicted by inflationary cosmologies _
, _ astrophys .
j. _ * 403 * ( 1993 ) l1 , [ http://xxx.lanl.gov/abs/astro-ph/9208001 [ arxiv : astro - ph/9208001 ] ] .
a. gangui , f. lucchin , s. matarrese , and s. mollerach , _ the three point correlation function of the cosmic microwave background in inflationary models _ , _ astrophys
. j. _ * 430 * ( 1994 ) 447457 , [ http://xxx.lanl.gov/abs/astro-ph/9312033 [ arxiv : astro - ph/9312033 ] ] . x. chen , m .- x .
huang , s. kachru , and g. shiu , _ observational signatures and non - gaussianities of general single field inflation _
, _ jcap _ * 0701 * ( 2007 ) 002 , [ http://xxx.lanl.gov/abs/hep-th/0605045 [ arxiv : hep - th/0605045 ] ] .
m. sasaki , h. suzuki , k. yamamoto , and j. yokoyama , _ superexpansionary divergence : breakdown of perturbative quantum field theory in space - time with accelerated expansion _ , _ class .
* 10 * ( 1993 ) l55l60 .
v. f. mukhanov , l. r. w. abramo , and r. h. brandenberger , _ on the back reaction problem for gravitational perturbations _
* 78 * ( 1997 ) 16241627 , [ http://xxx.lanl.gov/abs/gr-qc/9609026[arxiv:gr-qc/9609026 ] ] .
l. r. w. abramo , r. h. brandenberger , and v. f. mukhanov , _ the energy - momentum tensor for cosmological perturbations _ , _ phys .
* d56 * ( 1997 ) 32483257 , [ http://xxx.lanl.gov/abs/gr-qc/9704037[arxiv:gr-qc/9704037 ] ] .
t. prokopec , o. tornkvist , and r. p. woodard , _ one loop vacuum polarization in a locally de sitter background _ , _ ann .
* 303 * ( 2003 ) 251274 , [ http://xxx.lanl.gov/abs/gr-qc/0205130 [ arxiv : gr - qc/0205130 ] ] .
d. boyanovsky , h. j. de vega , and n. g. sanchez , _ quantum corrections to slow roll inflation and new scaling of superhorizon fluctuations _ , _ nucl .
_ * b747 * ( 2006 ) 2554 , [ http://xxx.lanl.gov/abs/astro-ph/0503669 [ arxiv : astro - ph/0503669 ] ] .
d. boyanovsky , h. j. de vega , and n. g. sanchez , _ quantum corrections to the inflaton potential and the power spectra from superhorizon modes and trace anomalies _ , _ phys .
* d72 * ( 2005 ) 103006 , [ http://xxx.lanl.gov/abs/astro-ph/0507596 [ arxiv : astro - ph/0507596 ] ] .
t. prokopec , n. c. tsamis , and r. p. woodard , _ two loop scalar bilinears for inflationary sqed _ , _ class .
* 24 * ( 2007 ) 201230 , [ http://xxx.lanl.gov/abs/gr-qc/0607094[arxiv:gr-qc/0607094 ] ] .
x. chen , m .- x .
huang , and g. shiu , _ the inflationary trispectrum for models with large non - gaussianities _ , _ phys .
_ * d74 * ( 2006 ) 121301 , [ http://xxx.lanl.gov/abs/hep-th/0610235 [ arxiv : hep - th/0610235 ] ] .
s. weinberg , _ quantum contributions to cosmological correlations .
ii : can these corrections become large ?
_ , _ phys .
* d74 * ( 2006 ) 023508 , [ http://xxx.lanl.gov/abs/hep-th/0605244[arxiv:hep-th/0605244 ] ] .
k. chaicherdsakul , _ quantum cosmological correlations in an inflating universe : can fermion and gauge fields loops give a scale free spectrum ? _ , _ phys .
rev . _ * d75 * ( 2007
) 063522 , [ http://xxx.lanl.gov/abs/hep-th/0611352[arxiv:hep-th/0611352 ] ] . p. adshead , r. easther , and e. a. lim , _ cosmology with many light scalar fields : stochastic inflation and loop corrections _ , _ phys .
rev . _ * d79 * ( 2009 ) 063504 , [ http://xxx.lanl.gov/abs/0809.4008 [ arxiv:0809.4008 ] ] .
y. l. dokshitzer , _ calculation of the structure functions for deep inelastic scattering and @xmath286 @xmath287 annihilation by perturbation theory in quantum chromodynamics _ , _ sov .
phys . jetp _ * 46 * ( 1977 ) 641653 .
gong and e. d. stewart , _ the density perturbation power spectrum to second - order corrections in the slow - roll expansion _ , _ phys .
lett . _ * b510 * ( 2001 ) 19 , [ http://xxx.lanl.gov/abs/astro-ph/0101225 [ arxiv : astro - ph/0101225 ] ] .
gong and e. d. stewart , _ the power spectrum for a multi - component inflaton to second - order corrections in the slow - roll expansion _ , _ phys .
_ * b538 * ( 2002 ) 213222 , [ http://xxx.lanl.gov/abs/astro-ph/0202098 [ arxiv : astro - ph/0202098 ] ] .
hwang and h. noh , _ third order perturbations of a zero - pressure cosmological medium : pure general relativistic nonlinear effects _ , _ phys .
* d72 * ( 2005 ) 044012 , [ http://xxx.lanl.gov/abs/gr-qc/0412129[arxiv:gr-qc/0412129 ] ] .
j. hwang and h. noh , _ third - order cosmological perturbations of zero - pressure multi - component fluids : pure general relativistic nonlinear effects _ , _ jcap _ * 0712 * ( 2007 ) 003 , [ http://xxx.lanl.gov/abs/0704.2086[arxiv:0704.2086 ] ] .
n. bartolo , s. matarrese , m. pietroni , a. riotto , and d. seery , _ on the physical significance of infra - red corrections to inflationary observables _ , _ jcap _ * 0801 * ( 2008 ) 015 , [ http://xxx.lanl.gov/abs/0711.4263[arxiv:0711.4263 ] ] .
k. enqvist , s. nurmi , d. podolsky , and g. i. rigopoulos , _ on the divergences of inflationary superhorizon perturbations _ , _ jcap _ * 0804 * ( 2008 ) 025 , [ http://xxx.lanl.gov/abs/0802.0395 [ arxiv:0802.0395 ] ] .
k. kohri , d. h. lyth , and c. a. valenzuela - toledo , _ on the generation of a non - gaussian curvature perturbation during preheating _ , _ jcap _ * 1002 * ( 2010 ) 023 , [ http://xxx.lanl.gov/abs/0904.0793 [ arxiv:0904.0793 ] ] .
s. hollands and r. m. wald , _ quantum field theory in curved spacetime , the operator product expansion , and dark energy _ , _ gen .
_ * 40 * ( 2008 ) 20512059 , [ http://xxx.lanl.gov/abs/0805.3419 [ arxiv:0805.3419 ] ] .
s. hollands and r. m. wald , _ existence of local covariant time ordered products of quantum fields in curved spacetime _
, _ commun .
phys . _ * 231 * ( 2002 ) 309345 , [ http://xxx.lanl.gov/abs/gr-qc/0111108 [ arxiv : gr - qc/0111108 ] ] .
s. hollands and r. m. wald , _ local wick polynomials and time ordered products of quantum fields in curved spacetime _ , _ commun .
* 223 * ( 2001 ) 289326 , [ http://xxx.lanl.gov/abs/gr-qc/0103074 [ arxiv : gr - qc/0103074 ] ] .
v. balasubramanian , p. kraus , and a. e. lawrence , _ bulk vs. boundary dynamics in anti - de sitter spacetime _ , _ phys .
_ * d59 * ( 1999 ) 046003 , [ http://xxx.lanl.gov/abs/hep-th/9805171 [ arxiv : hep - th/9805171 ] ] . c. gordon , d. wands , b. a. bassett , and r. maartens , _ adiabatic and entropy perturbations from inflation _ , _ phys
_ * d63 * ( 2001 ) 023506 , [ http://xxx.lanl.gov/abs/astro-ph/0009131 [ arxiv : astro - ph/0009131 ] ] .
m. sasaki and e. d. stewart , _ a general analytic formula for the spectral index of the density perturbations produced during inflation _ , _ prog .
* 95 * ( 1996 ) 7178 , [ http://xxx.lanl.gov/abs/astro-ph/9507001 [ arxiv : astro - ph/9507001 ] ] .
d. seery and j. c. hidalgo , _ non - gaussian corrections to the probability distribution of the curvature perturbation from inflation _ , _ jcap _ * 0607 * ( 2006 ) 008 , [ http://xxx.lanl.gov/abs/astro-ph/0604579 [ arxiv : astro - ph/0604579 ] ]
. k. kirsten and j. garriga , _ massless minimally coupled fields in de sitter space : o(4 ) symmetric states versus de sitter invariant vacuum _ , _ phys .
_ * d48 * ( 1993 ) 567577 , [ http://xxx.lanl.gov/abs/gr-qc/9305013[arxiv:gr-qc/9305013 ] ] .
a. higuchi and r. h. weeks , _ the physical graviton two - point function in de sitter spacetime with @xmath288 spatial sections _
, _ class .
grav . _ * 20 * ( 2003 ) 30053022 , [ http://xxx.lanl.gov/abs/gr-qc/0212031[arxiv:gr-qc/0212031 ] ] .
y. urakawa and k .-
cosmological density fluctuations in stochastic gravity formalism and linear analysis _
* d77 * ( 2008 ) 024013 , [ http://xxx.lanl.gov/abs/0710.5342 [ arxiv:0710.5342 ] ] .
y. urakawa and k .-
one - loop corrections to scalar and tensor perturbations during inflation in stochastic gravity _ , _ phys .
_ * d78 * ( 2008 ) 064004 , [ http://xxx.lanl.gov/abs/0801.0126 [ arxiv:0801.0126 ] ] .
p. creminelli , s. dubovsky , a. nicolis , l. senatore , and m. zaldarriaga , _ the phase transition to slow - roll eternal inflation _ , _ jhep _ * 09 * ( 2008 ) 036 , [ http://xxx.lanl.gov/abs/0802.1067 [ arxiv:0802.1067 ] ] .
a. a. starobinsky and j. yokoyama , _ equilibrium state of a self - interacting scalar field in the de sitter background _ , _ phys .
rev . _ * d50 * ( 1994 ) 63576368 , [ http://xxx.lanl.gov/abs/astro-ph/9407016 [ arxiv : astro - ph/9407016 ] ] . | in this article , i briefly review the status of infrared effects which occur when using inflationary models to calculate initial conditions for a subsequent hot , dense plasma phase .
three types of divergence have been identified in the literature : secular , `` time - dependent '' logarithms , which grow with time spent outside the horizon ; `` box - cutoff '' logarithms , which encode a dependence on the infrared cutoff when calculating in a finite - size box ; and `` quantum '' logarithms , which depend on the ratio of a scale characterizing new physics to the scale of whatever process is under consideration , and whose interpretation is the same as conventional field theory .
i review the calculations in which these divergences appear , and discuss the methods which have been developed to deal with them
. diags |
a good understanding of quantum chromodynamics ( qcd ) at finite temperature and baryon density is crucial for us to understand a wide range of physical phenomena .
for instance , to understand the evolution of the universe in the first few seconds , one needs the knowledge of qcd phase transition at temperature @xmath3mev and very small baryon density . on the other hand ,
understanding the physics of neutron stars requires the knowledge of qcd at high baryon density and very low temperature @xcite .
lattice simulation of qcd at finite temperature has been successfully performed in the past few decades ; however , no successful lattice simulation at high baryon density has been done due to the sign problem @xcite : the fermion determinant is not positively definite in presence of a nonzero baryon chemical potential @xmath0 .
we thus look for some special cases which have a positively definite fermion determinant .
one case is qcd at finite isospin chemical potential @xmath4 @xcite , where the ground state changes from a pion condensate to a bcs superfluid with quark - antiquark condensation with increasing isospin density .
another case is the qcd - like theories @xcite where quarks are in a real or pseudoreal representation of the gauge group , including two - color qcd with quarks in the fundamental representation and qcd with quarks in the adjoint representation . while these cases do not correspond to the real world , they can be simulated on the lattice and may give us some information of real qcd at finite baryon density .
for all these special cases , chiral perturbation theories predict a continuous quantum phase transition from the vacuum to the matter phase at baryon or isospin chemical potential equal to the pion mass , in contrast to real qcd where the phase transition takes place at @xmath0 approximately equal to the nucleon mass .
the resulting matter near the quantum phase transition is a dilute bose condensate of diquarks or pions with weakly repulsive interactions @xcite .
the equations of state and elementary excitations in such matter have been investigated many years ago by bogoliubov @xcite and lee , huang , and yang @xcite .
bose - einstein condensation ( bec ) phenomenon is believed to widely exist in dense matter , such as pions and kaons can condense in neutron star matter if the electron chemical potential exceeds the effective mass for pions and kaons @xcite .
however , the condensation of pions and kaons in neutron star matter is rather complicated due to the meson - nucleon interactions in dense nuclear medium . on the other hand , at asymptotically high density
, perturbative qcd calculations show that the ground state is a weakly coupled bcs superfluid with the condensation of overlapping cooper pairs @xcite .
it is interesting that the dense bcs superfluid and the dilute bose condensate have the same symmetry breaking pattern and thus are continued with one another . in condensed matter physics , this phenomenon was first discussed by eagles @xcite and leggett @xcite and is now called bec - bcs crossover .
it has been successfully realized using ultracold fermionic atoms in the past few years @xcite .
while the lattice simulations of two - color qcd at finite baryon chemical potential @xcite and qcd at finite isospin chemical potential @xcite have been successfully performed , we still ask for some effective models to link the physics of bose condensate and the bcs superfluidity . the chiral perturbation theories @xcite as well as the linear sigma models @xcite , which describe only the physics of bose condensate , do not meet our purpose .
the nambu jona - lasinio ( njl ) model @xcite with quarks as elementary blocks , which describes well the mechanism of chiral symmetry breaking and low energy phenomenology of the qcd vacuum , is generally believed to work at low and moderate temperatures and densities @xcite .
recently , this model has been used to describe the superfluid transition at finite chemical potentials @xcite for the special cases we are interested in this paper .
one finds that the critical chemical potential for the superfluid transition predicted by the njl model is indeed equal to the pion mass @xcite , and the chiral and diquark condensates obtained from the mean - field calculation agree with the results from lattice simulations and chiral perturbation theories near the quantum phase transition @xcite .
the njl model also predicts a bec - bcs crossover when the chemical potential increases @xcite .
a natural problem arises : how can the fermionic njl model describe the weakly interacting bose condensate near the quantum phase transition ?
in fact , we do not know how the repulsive interactions among diquarks or mesons enter in the pure mean - field calculations @xcite . in this paper
, we will focus on this problem and show that the repulsive interaction is indeed properly included even in the mean - field calculations .
this phenomenon is in fact analogous to the bcs description of the molecular condensation in strongly interacting fermi gases studied by leggett many years ago @xcite .
fermionic models have been used to describe the bec - bcs crossover in cold fermi gases by the cold atom community .
recently , it has been shown that we can recover the equation of state of the dilute bose condensate with correct boson - boson scattering length in the strong coupling limit , including the lee - huang - yang correction by considering the beyond - mean - field corrections @xcite . in appendix
[ app1 ] , we give a summary of the many - body theoretical approach in cold atoms , which is useful for us to understand the theoretical approach and the results of this paper . in this paper , using two - color two - flavor qcd as an example and following the theoretical approach of the bec - bcs crossover in cold fermi gases @xcite , we examine how the njl model describes the weakly interacting bose condensate and the bec - bcs crossover . near the quantum phase transition point @xmath1
, we perform a ginzburg - landau expansion of the effective potential at the mean - field level , and show that the ginzburg - landau free energy is essentially the gross - pitaevskii free energy describing weakly interacting bose condensates via a proper redefinition of the condensate wave function . as a by - product , we obtain a diquark - diquark scattering length @xmath5 ( @xmath6 is the pion decay constant ) characterizing the repulsive interaction between the diquarks , which recovers the tree - level result predicted by chiral lagrangian @xcite .
we also show analytically that the goldstone mode takes the same dispersion as the bogoliubov excitation in weakly interacting bose condensates , which gives a diquark - diquark scattering length identical to that in the gross - pitaevskii free energy .
the mixing between the sigma meson and diquarks plays an important role in recovering the bogoliubov excitation .
the results of in - medium chiral and diquark condensates predicted by chiral perturbation theory are analytically recovered . at high density
, we find the superfluid matter undergoes a bec - bcs crossover at @xmath7 with @xmath8 being the mass of the sigma meson . at @xmath9 , we find that the chiral symmetry is approximated restored and the spectra of pions and sigma meson become nearly degenerated .
well above the chemical potential of chiral symmetry restoration , the degenerate pions and sigma
meson undergo a mott transition , where they become unstable resonances . because of the spontaneous breaking of baryon number symmetry
, mesons can decay into quark pairs in the superfluid medium at nonzero momentum .
the beyond - mean - field corrections are studied .
the thermodynamic potential including the gaussian fluctuations is derived .
it is shown that the vacuum state @xmath10 is thermodynamically consistent in the gaussian approximation , i.e. , all thermodynamic quantities keep vanishing in the regime @xmath10 even though the beyond - mean - field corrections are included .
near the quantum phase transition point , we expand the fluctuation contribution to the thermodynamic potential in powers of the superfluid order parameter . to leading order
, the beyond - mean - field correction is quartic and its effect is to renormalize the diquark - diquark scattering length .
the correction to the mean - field result is shown to be proportional to @xmath11 .
thus , our theoretical approach provides a new way to calculate the diquark - diquark or meson - meson scattering lengths in the njl model beyond the mean - field approximation .
we also find that we can obtain a correct transition temperature of bose condensation in the dilute limit , including the beyond - mean - field corrections .
the paper is organized as follows : in sec . [ s2 ] , we derive the general effective action of the two - color njl model at finite temperature and density , and determine the model parameters via the vacuum phenomenology . in sec .
[ s3 ] , we investigate the properties of dilute bose condensate near the quantum phase transition at the mean - field level . in sec .
[ s5 ] , the properties of matter at high density are discussed .
beyond - mean - field corrections are studied in sec .
we summarize in sec .
natural units are used throughout .
without loss of generality , we study in this paper two - color qcd ( the number of colors @xmath12 ) at finite baryon chemical potential @xmath0 . for vanishing current quark mass @xmath13 , two - color qcd possesses an enlarged flavor symmetry su@xmath14 [ @xmath15 is the number of flavors ] , the so - called pauli - gursey symmetry which connects quarks and antiquarks @xcite . for @xmath16 , the flavor symmetry su@xmath17
is spontaneously broken down to sp@xmath17 driven by a nonzero quark condensate @xmath18 and there arise five goldstone bosons : three pions and two scalar diquarks . for nonvanishing current quark mass , the flavor symmetry is explicitly broken , resulting in five pseudo - goldstone bosons with a small degenerate mass @xmath2 . at finite baryon chemical potential @xmath0 , the flavor symmetry su@xmath14 is explicitly broken down to su@xmath19su@xmath20u@xmath21 .
further , a nonzero diquark condensate @xmath22 can form at large enough chemical potentials and breaks spontaneously the u@xmath21 symmetry . in two - color qcd ,
the scalar diquarks are in fact the lightest baryons , " and we expect a baryon superfluid phase with @xmath23 for @xmath24 . to construct a njl model for two - color two - flavor qcd with the above flavor symmetry , we consider a contact current - current interaction @xmath25 where @xmath26 ( @xmath27 ) are the generators of color su@xmath28 and @xmath29 is a phenomenological coupling constant .
after the fierz transformation we can obtain an effective njl lagrangian density with scalar mesons and color singlet scalar diquarks @xcite , @xmath30,\end{aligned}\ ] ] where @xmath31 and @xmath32 are the charge conjugate spinors with @xmath33 and @xmath34 ( @xmath35 ) are the pauli matrices in the flavor space .
the four - fermion coupling constants for the scalar mesons and diquarks are the same , @xmath36 @xcite , which ensures the enlarged flavor symmetry su@xmath14 of two - color qcd in the chiral limit @xmath37 .
one can show explicitly that there are five goldstone bosons ( three pions and two diquarks ) driven by a nonzero quark condensate @xmath18 . with explicit chiral symmetry broken @xmath38 , pions and diquarks
are also degenerate , and their mass @xmath2 can be determined via the standard method for the njl model @xcite .
the partition function of the two - color njl model ( [ njl ] ) at finite temperature @xmath39 and baryon chemical potential @xmath0 is @xmath40[dq]\exp\left[\int dx\left({\cal l}_{\text{njl}}+\frac{\mu_{\text b}}{2}\bar{q}\gamma_{0}q\right)\right],\end{aligned}\ ] ] where we adopt the finite temperature formalism with @xmath41 , @xmath42 , and @xmath43 .
the partition function can be bosonized after introducing the auxiliary boson fields @xmath44 for mesons and @xmath45 for diquarks . with the help of the nambu - gorkov representation @xmath46
, the partition function can be written as @xmath47[d\psi][d\sigma][d\mbox{\boldmath{$\pi$}}][d\phi^\dagger][d\phi]\exp\left(-{\cal a_{\text{eff}}}\right),\end{aligned}\ ] ] where the action @xmath48 is given by @xmath49 with the inverse quark propagator defined as @xmath50 here @xmath51 . after integrating out the quarks
, we can reduce the partition function to @xmath52[d\mbox{\boldmath{$\pi$}}][d\phi^\dagger][d\phi]\exp\big\{-{\cal s}_{\text{eff}}[\sigma,\mbox{\boldmath{$\pi$}},\phi^\dagger,\phi]\big\}$ ] , where the bosonized effective action @xmath53 is given by @xmath54=\int dx\frac{\sigma^2(x)+\mbox{\boldmath{$\pi$}}^2(x)+|\phi(x)|^2}{4g}-\frac{1}{2}\text{tr}\ln{\bf g}^{-1}(x , x^\prime).\end{aligned}\ ] ] here the trace @xmath55 is taken over color , flavor , spin , nambu - gorkov and coordinate ( @xmath56 and @xmath57 ) spaces .
the thermodynamic potential density of the system is given by @xmath58 .
the effective action @xmath53 as well as the thermodynamic potential @xmath59 can not be evaluated exactly in our @xmath60 dimensional case . in this work ,
we firstly consider the saddle point approximation , i.e. , the mean - field approximation .
then we investigate the fluctuations around the mean field . + _
( i)mean - field approximation .
_ in this approximation , all bosonic auxiliary fields are replaced by their expectation values . to this end ,
we write @xmath61 , @xmath62 and set @xmath63 .
while @xmath64 can be set to be real , we do not do this first in our derivations .
we will show in the following that all physical results depend only on @xmath65 .
the zeroth order or mean - field effective action reads @xmath66.\ ] ] here and in the following @xmath67 with @xmath68 being the fermion matsubara frequency , and @xmath69 with @xmath70 .
the inverse of the nambu - gorkov quark propagator @xmath71 is given by @xmath72 with the effective quark dirac mass @xmath73 .
the mean - field thermodynamic potential @xmath74 can be evaluated as @xmath75\end{aligned}\ ] ] with the definitions of the function @xmath76 and the bcs - like quasiparticle dispersions @xmath77 where @xmath78 .
the signs @xmath79 correspond to quasiquark and quasi - antiquark excitations , respectively .
the integral over the quark momentum @xmath80 is divergent at large @xmath81 , and some regularization scheme should be adopted . in this paper , we employ a hard three - momentum cutoff @xmath82 .
the physical values of the variational parameters @xmath83 ( or @xmath84 ) and @xmath64 should be determined by the saddle point condition @xmath85}{\delta \upsilon}=0,\ \ \ \ \
\frac{\delta{\cal s}_{\text{eff}}^{(0)}[\upsilon,\delta]}{\delta \delta}=0,\end{aligned}\ ] ] which minimizes the mean - field effective action @xmath86 .
one can show that the saddle point condition is equivalent to the following green function relations @xmath87\ , \end{aligned}\ ] ] where the matrix elements of @xmath88 are explicitly given by @xmath89 with the help of the massive energy projectors @xcite @xmath90\ .\ ] ] here we have defined the notation @xmath91 for convenience .
( ii)derivative expansion .
_ next , we consider the fluctuations around the mean field , corresponding to the bosonic collective excitations . making the field shifts for the auxiliary fields , @xmath92 we can express the total effective action as @xmath93}.\end{aligned}\ ] ] here @xmath94 is the fourier transformation of @xmath95 , and @xmath96 is defined as @xmath97 with the help of the derivative expansion @xmath98}=\sum_{n=1}^\infty\frac{(-1)^{n+1}}{n}\text{tr}[{\cal g}\sigma]^n,\end{aligned}\ ] ] we can calculate the effective action in powers of the fluctuations @xmath99 .
the first order effective action @xmath100 which includes linear terms of the fluctuations should vanish exactly , since the expectation value of the fluctuations should be exactly zero .
in fact , @xmath100 can be evaluated as @xmath101\sigma(x)\nonumber\\ & + & \frac{1}{2}\text{tr}\left[i\gamma_5\left({\cal g}_{11}\mbox{\boldmath{$\tau$}}+{\cal g}_{22}\mbox{\boldmath{$\tau$}}^{\text t}\right)\right]\cdot\mbox{\boldmath{$\pi$}}(x)\nonumber\\ & + & \left[\frac{\delta}{2g}+\frac{1}{2}\text{tr}\left(i\gamma_5\tau_2 t_2{\cal g}_{12}\right)\right]\phi^\dagger(x)\nonumber\\ & + & \left[\frac{\delta^\dagger}{2g}+\frac{1}{2}\text{tr}\left(i\gamma_5\tau_2 t_2{\cal g}_{21}\right)\right]\phi(x)\bigg\}.\end{aligned}\ ] ] we observe that the coefficients of @xmath102 is automatically zero after taking the trace in dirac spin space .
the coefficients of @xmath103 and @xmath104 vanish once the quark propagator takes the mean - field form and @xmath105 take the physical values satisfying the saddle point condition .
thus , in the present approach , the saddle point condition plays a crucial role in having a vanishing linear term in the expansion .
the quadratic term @xmath106 corresponds to the gaussian fluctuations .
it reads @xmath107.\end{aligned}\ ] ] for the convenience of our investigation in the following , we will use the form of @xmath106 in the momentum space .
after the fourier transformation , it can be written as @xmath108\bigg\}.\end{aligned}\ ] ] where @xmath109 with @xmath110 being the boson matsubara frequency and @xmath111 . here
@xmath112 is the fourier transformation of the field @xmath113 , and @xmath114 is defined as @xcite @xmath115 + _ ( iii)gaussian fluctuations . _ after taking the trace in nambu - gorkov space , we find that @xmath106 can be written in the following bilinear form @xmath116 the matrix @xmath117 takes the following nondiagonal form @xmath118 the polarization functions @xmath119 ( @xmath120 ) are one - loop susceptibilities composed of the matrix elements the nambu - gorkov quark propagator , and can be expressed as @xmath121,\ \ \ \
\pi_{22}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{11}(k)\gamma{\cal g}_{22}(p)\gamma\right],\nonumber\\ & & \pi_{12}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{12}(k)\gamma{\cal g}_{12}(p)\gamma\right],\ \ \ \
\pi_{21}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{21}(k)\gamma{\cal g}_{21}(p)\gamma\right],\nonumber\\ & & \pi_{33}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{11}(k){\cal g}_{11}(p)+{\cal g}_{22}(k){\cal g}_{22}(p)+{\cal g}_{12}(k){\cal g}_{21}(p)+{\cal g}_{21}(k){\cal g}_{12}(p)\right],\nonumber\\ & & \pi_{13}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{12}(k)\gamma{\cal g}_{11}(p)+{\cal g}_{22}(k)\gamma{\cal g}_{12}(p)\right],\ \ \
\pi_{31}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{21}(k){\cal g}_{11}(p)\gamma+{\cal g}_{22}(k){\cal g}_{21}(p)\gamma\right],\nonumber\\ & & \pi_{23}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{11}(k)\gamma{\cal g}_{21}(p)+{\cal g}_{21}(k)\gamma{\cal g}_{22}(p)\right],\ \ \
\pi_{32}(q)=\frac{1}{2}\sum_k\text{tr}\left[{\cal g}_{11}(k){\cal g}_{12}(p)\gamma+{\cal g}_{12}(k){\cal g}_{22}(p)\gamma\right],\end{aligned}\ ] ] where @xmath122 , @xmath123 and the trace is taken over color , flavor and spin spaces .
using the fact that @xmath124 and @xmath125 , we can easily show that @xmath126 therefore , only five of the polarization functions are independent . at @xmath127 , their explicit form is shown in appendix [ app2 ] . for general case , we can show that @xmath128 and @xmath129 .
thus , in the normal phase where @xmath130 , the matrix @xmath117 recovers the diagonal form .
the off - diagonal elements @xmath131 and @xmath132 represents the mixing between the sigma meson and diquarks . at large chemical potentials where the chiral symmetry is approximately restored , @xmath133 , this mixing can be safely neglected .
on the other hand , the matrix @xmath134 of the pion sector is diagonal and proportional to the identity matrix , i.e. , @xmath135 , \ \
\ \text{i , j}=1,2,3.\ ] ] this means pions are eigen mesonic excitations even in the superfluid phase . the polarization function @xmath136 is given by @xmath137.\end{aligned}\ ] ] its explicit form at @xmath127 is shown in appendix [ app2 ] .
we find that @xmath138 and @xmath139 is different only to a term proportional to @xmath140 .
thus , at high density where @xmath141 , the spectra of pions and sigma meson become nearly degenerate which represents the approximate restoration of chiral symmetry . + _ ( iv)goldstone s theorem . _ the u@xmath21 baryon number symmetry is spontaneously broken by the nonzero diquark condensate @xmath22 in the superfluid phase , resulting in one goldstone boson . in our model
, this is ensured by the condition @xmath142 . from the explicit form of the polarization functions shown in appendix [ app2 ]
, we find that this condition holds if and only if the saddle point condition ( [ saddle ] ) for @xmath84 and @xmath64 is satisfied .
we thus emphasize that in our theoretical framework , the condensates @xmath84 and @xmath64 should be determined by the saddle point condition , and the beyond - mean - field corrections are possible only through the thermodynamics , i.e. , the equations of state .
for a better understanding our derivation in the following , it is useful to review the vacuum state at @xmath143 . in the vacuum ,
it is evident that @xmath130 and the mean - field effective potential @xmath144 can be evaluated as @xmath145 the physical value of @xmath83 , denoted by @xmath146 , satisfies the saddle point condition @xmath147 and minimizes @xmath144 .
the meson and diquark excitations can be obtained from @xmath148 , which in the vacuum can be expressed as @xmath149,\end{aligned}\ ] ] where @xmath150 are the real and imaginary parts of @xmath151 , respectively .
the inverse propagators in vacuum can be expressed in a symmetrical form @xcite @xmath152 where @xmath153 , @xmath154 , and the function @xmath155 is defined as @xmath156 with @xmath157 . keeping in mind that @xmath146 satisfies the saddle point condition , we find that the pions and diquarks are nambu - goldstone bosons in the chiral limit , corresponding to the symmetry breaking pattern su@xmath158sp@xmath17 . using the gap equation of @xmath146 , we find that the masses of mesons and diquarks can be determined by the equation @xmath159 since the @xmath160 dependence of the function @xmath155 is very weak , we find @xmath161 and @xmath162 .
since pions and diquarks are deep bound states , their propagators can be well approximated by @xmath163 with @xmath164 .
the pion decay constant @xmath6 can be determined by the matrix element of the axial current , @xmath165\nonumber\\ & = & 2n_cn_fg_{\pi qq}m_*q_\mu i(q^2)\delta_{\text{ij}}.\end{aligned}\ ] ] here @xmath166 .
thus , the pion decay constant can be expressed as @xmath167 finally , together with ( [ mesonmass ] ) and ( [ piondecay ] ) , we recover the well - known gell - mann oakes renner relation @xmath168 .
.model parameters ( 3-momentum cutoff @xmath82 , coupling constant @xmath169 , and current quark mass @xmath13 ) and related quantities ( quark condensate @xmath170 , constituent quark mass @xmath146 and pion mass @xmath2 ) for the two - flavor two - color njl model ( [ njl ] ) . the pion decay constant is fixed to be @xmath171 mev . [ cols="^,^,^,^,^,^,^ " , ] on the other hand , we find from the explicit forms of the meson propagators in appendix [ app2 ] that the decay process @xmath172 is also possible at @xmath173 ( even though @xmath174 is small ) due to the presence of superfluidity .
thus , we have another unusual mott transition in the superfluid phase .
notice that this process is not in contradiction to the baryon number conservation law , since the u@xmath21 baryon number symmetry is spontaneously broken in the superfluid phase .
quantitatively , this transition occurs when the meson mass becomes larger than the two - particle continuum @xmath175 for the decay process @xmath172 at @xmath176 . in this case
, we have @xmath177 the two - particle continuum @xmath175 is also shown in fig.[fig9 ] .
we find that the unusual mott transition does occur at another chemical potential @xmath178 which is also sensitive to the value of @xmath146 .
the values of @xmath179 for the four model parameter sets are also shown in table.[mottmu ] . for reasonable model parameter sets ,
this value is in the range @xmath180 .
this process can also occur in the 2sc phase of quark matter in the @xmath181 case @xcite . in the 2sc phase ,
the symmetry breaking pattern is su@xmath182u@xmath183su@xmath184 where the generator of the residue baryon number symmetry @xmath185 is @xmath186 corresponding to the unpaired blue quarks .
thus the baryon number symmetry for the paired red and green quarks are broken and our results can be applied . to show this explicitly , we write down the explicit form of the polarization function for pions in the 2sc phase @xcite @xmath187,\ ] ] where @xmath188 is the propagator for the unpaired blue quarks . here
@xmath189 is given by ( [ pionpo ] ) ( the effective quarks mass @xmath83 and the pairing gap @xmath64 should be given by the @xmath181 case of course ) and corresponds to the contribution from the paired red and green sectors .
the second term is the contribution from the unpaired blue quarks .
therefore , the unusual decay process is only available for the paired quarks .
the investigations in sec . [ s3 ] and [ s4 ] are restricted in the mean - field approximation , even though the bosonic collective excitations are studied . in this section
, we will include the gaussian fluctuations in the thermodynamic potential , and thus really go beyond the mean field .
the scheme of going beyond the mean field is somewhat like those done in the study of finite temperature thermodynamics of the njl model @xcite ; however , in this paper we will focus on the beyond - mean - field corrections at zero temperature , i.e. , the pure quantum fluctuations .
we will first derive the thermodynamic potential beyond the mean field which is valid at arbitrary chemical potential and temperature , and then briefly discuss the beyond - mean - field corrections near the quantum phase transition .
the numerical calculations are deferred for future studies . in the gaussian approximation , the partition function can be expressed as @xmath190[d\mbox{\boldmath{$\pi$}}][d\phi^\dagger][d\phi]\exp\big(-{\cal s}_{\text{eff}}^{(2)}\big).\end{aligned}\ ] ] integrating out the gaussian fluctuations , we can express the total thermodynamic potential as @xmath191 where the contribution from the gaussian fluctuations can be written as @xmath192.\end{aligned}\ ] ] however , there is a problem with the above expression , since it is actually ill - defined : the sum over the boson matsubara frequency is divergent and we need appropriate convergent factors to make it meaningful . in the simpler case without superfluidity ,
the convergent factor is simply given by @xmath193 @xcite . in our case , the situation is somewhat different due to the introduction of the nambu - gorkov spinors .
keep in mind that in the equal time limit , there are additional factors @xmath194 for @xmath195 and @xmath196 for @xmath197 .
therefore , to get the proper convergent factors for @xmath198 , we should keep these factors when we make the sum over the fermion matsubara frequency @xmath199 in evaluating the polarization functions @xmath119 and @xmath138 .
the problem in the expression of @xmath198 is thus from the opposite convergent factors for @xmath200 and @xmath201 . from the above arguments
, we find that there is a factor @xmath193 for @xmath200 and @xmath202 for @xmath201 .
keep in mind that the matsubara sum @xmath203 is converted to a standard contour integral ( @xmath204 ) . the convergence for @xmath205 is automatically guaranteed by the bose distribution function @xmath206
, we thus should treat only the problem for @xmath207 . to this end
, we write the first term of @xmath198 as @xmath208.\end{aligned}\ ] ] using the fact that @xmath209 , we obtain @xmath210e^{i\nu_m 0^+}.\end{aligned}\ ] ] therefore , the well - defined form of @xmath198 is given by the above formula together with the other term @xmath211 associated with a factor @xmath193 .
the matsubara sum can be written as the contour integral via the theorem @xmath212 , where @xmath213 runs on either side of the imaginary @xmath214 axis , enclosing it counterclockwise . distorting the contour to run above and below the real axis ,
we obtain @xmath215,\end{aligned}\ ] ] where the scattering phases are defined as @xmath216.\end{aligned}\ ] ] keep in mind the pressure of the vacuum should be zero , the physical thermodynamic potential at finite temperature and chemical potential should be defined as @xmath217 as we have shown in the mean - field theory , at @xmath127 , the vacuum state is restricted in the region @xmath10 . in this region , all thermodynamic quantities should keep zero , no matter how large the value of @xmath0 is . while this should be an obvious physical conclusion , it is important to check whether our beyond - mean - field theory satisfies this condition .
notice that the physical thermodynamic potential is defined as @xmath218 , we therefore should prove that the thermodynamic potential @xmath219 keeps a constant in the region @xmath10 .
for the mean - field part @xmath220 , the proof is quite easy . because of the fact that @xmath221 , the solution for @xmath83 is always given by @xmath222 .
thus @xmath220 keeps its value at @xmath223 in the region @xmath224 .
now we turn to the complicated part @xmath198 .
since @xmath130 , all the off - diagonal elements of @xmath117 vanishes , and @xmath198 is reduced to @xmath225e^{i\nu_m 0^+}\nonumber\\ & + & \frac{3}{2}\sum_q\ln\left[\frac{1}{2g}+\pi_{\pi}(q)\right]e^{i\nu_m 0^+}\nonumber\\ & + & \sum_q\ln\left[\frac{1}{4g}+\pi_{\text d}(q)\right]e^{i\nu_m 0^+},\end{aligned}\ ] ] where @xmath226 and we should set @xmath130 and @xmath222 in evaluating the polarization functions .
first , we can easily show that the contributions from the sigma meson and pions do not have explicit @xmath0 dependence and thus keep the same values as those at @xmath223 .
in fact , since the effective quark mass @xmath83 keeps its vacuum value @xmath146 guaranteed by the mean - field part , all the @xmath0 dependence in @xmath227 is included in the fermi distribution functions @xmath228 . since @xmath229 , they vanish automatically at @xmath127 .
in fact , from the explicit expressions for @xmath227 in appendix [ app2 ] , we can check that there is no @xmath0 independence in @xmath227 .
the diquark contribution , however , has an explicit @xmath230 dependence through the combination @xmath231 in the polarization function @xmath232 .
the diquark contribution ( at @xmath127 ) can be written as @xmath233.\end{aligned}\ ] ] making a shift @xmath234 , and noticing that fact @xmath235 , we obtain @xmath236 to show the above quantity is in fact @xmath0 independent , we separate it into a pole part and a continuum part . there is a well - defined two - particle continuum @xmath237 for pions at arbitrary momentum @xmath238 , @xmath239 the pion propagator has two symmetric poles @xmath240 when @xmath238 satisfies @xmath241 . thus in the region
@xmath242 , the scattering phase @xmath243 can be analytically evaluated as @xmath244.\end{aligned}\ ] ] since @xmath245 , the thermodynamic potential @xmath246 can be separated as @xmath247-\sum_{\bf q}\int_{-\infty}^{-e_c({\bf q})}\frac{d\omega}{\pi}\delta_{\pi}(\omega,{\bf q}),\end{aligned}\ ] ] which is indeed @xmath0 independent .
notice that in the first term the integral over @xmath238 is restricted in the region @xmath248 where @xmath249 is defined as @xmath250 . in conclusion
, we have shown that the thermodynamic potential @xmath59 in the gaussian approximation keeps a constant in the vacuum state , i.e. , at @xmath10 and at @xmath127 .
all other thermodynamic quantities such as the baryon number density keep zero in the vacuum .
the subtraction term @xmath251 in the gaussian approximation can be expressed as @xmath252\nonumber\\ & & \ \ \
-\sum_{\bf q}\int_{-\infty}^{-e_c({\bf q})}\frac{d\omega}{2\pi}\left[\delta_\sigma(\omega,{\bf q})+5\delta_{\pi}(\omega,{\bf q})\right].\end{aligned}\ ] ] now we consider the beyond - mean - field corrections near the quantum phase transition point @xmath1 .
notice that the effective quark mass @xmath83 and the diquark condensate @xmath64 are determined at the mean - field level , and the beyond - mean - field corrections are possible only through the equations of state .
formally , the gaussian contribution to the thermodynamic potential @xmath198 is a function of @xmath253 and @xmath254 , i.e. , @xmath255 . in the superfluid phase ,
the total baryon density including the gaussian contribution can be evaluated as @xmath256 where the mean - field part is simply given by @xmath257 and the gaussian contribution can be expressed as @xmath258 the physical values of @xmath83 and @xmath65 should be determined by their mean - field gap equations . in fact , from the gap equations @xmath259 and @xmath260 , we obtain @xmath261 thus , we can obtain the derivatives @xmath262 and @xmath263 analytically . finally , @xmath264 is a continuous function of @xmath0 guaranteed by the properties of second order phase transition , and we have @xmath265 .
next we focus on the beyond - mean - field corrections near the quantum phase transition .
since the diquark condensate @xmath64 is vanishingly small , we can expand the gaussian part @xmath198 in powers of @xmath65 .
notice that @xmath0 and @xmath83 can be evaluated as functions of @xmath65 from the ginzburg - landau potential and mean - field gap equations .
thus to order @xmath266 , the expansion takes the form @xmath267 where the expansion coefficient @xmath268 is defined as @xmath269 using the definition of @xmath270 , we find that @xmath268 can be related to @xmath270 by @xmath271 thus , the coefficient @xmath268 vanishes , and the leading order of the expansion should be @xmath272 . as shown above , to leading order , the expansion of @xmath198 can be formally expressed as @xmath273 the method to derive the exact expression of the numerical factor @xmath274 is shown in appendix [ app3 ] .
notice that the factor @xmath274 is in fact @xmath0 independent , thus the total baryon density to leading order is @xmath275 near the quantum phase transition point , the mean - field contribution is @xmath276 from the gross - pitaevskii free energy .
the last term can be evaluated using the analytical result @xmath277 on the other hand , the total pressure @xmath278 can be expressed as @xmath279 thus we find that the leading order quantum corrections are totally included in the numerical factor @xmath274 . setting @xmath280
, we recover the mean - field results obtained in sec .
[ s3 ] . including the quantum fluctuations , the equations of state shown in ( [ eos ] )
are modified to be @xmath281 this means , to leading order , the effect of quantum fluctuations is giving a correction to the diquark - diquark scattering length .
the renormalized scattering length is @xmath282 generally , we have @xmath283 and the renormalized scattering length is smaller than the mean - field result .
an exact calculation of the numerical factor @xmath274 can be performed using the method shown in appendix [ app3 ] .
however , this needs huge numerical power and we defer it to future work @xcite . in this paper
we will give an analytical estimation of @xmath274 based on the fact that the quantum fluctuations are dominated by the gapless goldstone mode . to this end , we approximate the gaussian contribution @xmath198 as @xmath284,\end{aligned}\ ] ] where @xmath285 is given by ( [ dipro ] ) and can be approximated by ( [ diproa ] ) . subtracting the value of @xmath198 at @xmath1 with @xmath130 , and using the result @xmath286 from the gross - pitaevskii equation , we find that @xmath274 can be evaluated as @xmath287 where the numerical factors @xmath288 and @xmath289 are given by @xmath290 ^ 2}.\end{aligned}\ ] ] here the dimensionless notations @xmath291 and @xmath292 are defined as @xmath293 and @xmath294 respectively .
notice that the integral over @xmath292 is divergent and hence such an estimation has no prediction power due to the fact that the njl model is nonrenormalizable .
however , regardless of the numerical factor @xmath295 , we find that @xmath296 .
thus , the correction should be small in the nonlinear sigma model limit @xmath297 .
while the effect of the gaussian fluctuations at zero temperature is to give a small correction to the diquark - diquark scattering length and the equations of state , it can be significant at finite temperature .
in fact , as the temperature approaches the critical value of superfluidity , the gaussian fluctuations should dominate . in this part
, we will show that to get a correct critical temperature in terms of the baryon density @xmath298 , we must go beyond the mean field .
the situation is analogous to the nozieres
schmitt - rink treatment of molecular condensation in strongly interacting fermi gases @xcite .
the transition temperature @xmath299 is determined by the thouless criterion @xmath300 which can be shown to be consistent with the saddle point condition @xmath301 .
its explicit form is a bcs - type gap equation @xmath302 meanwhile , the dynamic quark mass @xmath83 satisfies the mean - field gap equation @xmath303 to obtain the transition temperature as a function of @xmath298 , we need the so - called number equation given by @xmath304 , which includes both the mean - field contribution @xmath305 $ ] and the gaussian contribution @xmath306 . at the transition temperature
where @xmath130 , @xmath198 can be expressed as @xmath307,\end{aligned}\ ] ] where the scattering phases are defined as @xmath308 $ ] for the diquarks , @xmath309 $ ] for the sigma meson and @xmath310 $ ] for the pions .
obviously , the polarization functions should take their forms at finite temperature in the normal phase .
the transition temperature @xmath299 at arbitrary baryon number density @xmath298 can be determined numerically via solving simultaneously the gap and number equations . however , in the dilute limit @xmath311 which we are interested in this section , analytical result can be achieved .
keep in mind that @xmath312 when @xmath313 , we find that the fermi distribution functions @xmath314 vanish exponentially ( since @xmath315 ) and we obtain @xmath1 and @xmath222 from the gap eqs .
( [ tc1 ] ) and ( [ tc2 ] ) , respectively .
meanwhile the mean - field contribution of the density @xmath316 can be neglected and the total density @xmath298 is thus dominated by the gaussian part @xmath270 .
when @xmath317 we can show that @xmath318 and @xmath319 are independent of @xmath0 , and the number equation is reduced to @xmath320 since @xmath321 , the inverse diquark propagator can be reduced to @xmath322 in ( [ dipro ] ) .
thus the scattering phase @xmath323 can be well approximated by @xmath324 $ ] with @xmath325 .
therefore , the number equation can be further reduced to the well - known equation for ideal bose - einstein condensation , @xmath326\bigg|_{\mu_{\text b}=m_\pi}.\end{aligned}\ ] ] since the above equation is valid only in the low density limit @xmath313 , the critical temperature is thus given by the nonrelativistic result @xmath327^{2/3}.\end{aligned}\ ] ] at finite density but @xmath328 , there exists a correction to @xmath299 which is proportional to @xmath329@xcite .
such a correction is hard to handle analytically in our model since we should consider simultaneously the corrections to @xmath83 and @xmath0 , as well as the contribution from the sigma meson and pions .
in summary , we have examined the njl model description of weakly interacting bose condensate and bec - bcs crossover in qcd - like theories at finite baryon density .
our main conclusions are as follows : + ( 1)near the quantum phase transition point @xmath1 , we have performed a ginzburg - landau expansion of the effective potential . at the mean - field level ,
the ginzburg - landau free energy is essentially the gross - pitaevskii free energy describing weakly repulsive bose condensates after a proper redefinition of the condensate wave function .
the obtained diquark - diquark scattering length reads @xmath330 , which recovers the tree - level result predicted by chiral lagrangian .
+ ( 2)we have analytically shown that the goldstone mode near the quantum phase transition point takes the same dispersion as the bogoliubov excitation in weakly interacting bose condensates , which gives a diquark - diquark scattering length identical to that in the gross - pitaevskii free energy .
the mixing between the sigma meson and the diquarks plays an important role in recovering the bogoliubov dispersion .
+ ( 3)the results of baryon number density and in - medium chiral and diquark condensates predicted by chiral perturbation theory are analytically recovered near the quantum phase transition point in the njl model .
+ ( 4)at high density , the superfluid matter undergoes a bec - bcs crossover at @xmath331 . at @xmath9 , the chiral symmetry is approximated restored and the spectra of pions and sigma meson become nearly degenerate .
well above the chemical potential of chiral symmetry restoration , the degenerate pions and sigma meson undergo a mott transition , where they become unstable resonances .
because of the spontaneous breaking of baryon number symmetry , mesons can decay into quark pairs in the superfluid medium at nonzero momentum
. + ( 5)the general theoretical framework of the thermodynamics beyond the mean field is established .
it is shown that the vacuum state in the region @xmath10 is thermodynamically consistent in the gaussian approximation , i.e. , all thermodynamic quantities keep vanishing for @xmath10 even though the gaussian fluctuations are included .
+ ( 6)near the quantum phase transition point , we find that the effect of the leading order beyond - mean - field correction is to renormalize the diquark - diquark scattering length .
the correction to the mean - field result is estimated to be proportional to @xmath11 .
our theoretical approach provides a new way to calculate the diquark - diquark or meson - meson scattering lengths in the njl model beyond the mean - field approximation .
we also find that we can obtain a correct transition temperature of bose condensation in the dilute limit once the beyond - mean - field corrections are included .
our studies can be generalized to describe pion condensation at finite isospin chemical potential @xmath4 @xcite and kaon condensation at finite strangeness chemical potential @xmath332 @xcite . in the njl model ,
pion condensation is shown to occur at @xmath333 when @xmath334 , and kaon is shown to condense at @xmath335 when @xmath336 @xcite .
the generalization to pion condensation is straightforward .
the obtained ginzburg - landau and gross - pitaevskii free energies are the same as those derived in this paper , if we replace @xmath337 .
the results are valid both for @xmath12 and @xmath181 cases . at the mean - field level ,
the results for diquark condensation at @xmath12 and pion condensation at @xmath181 are formally identical in the njl model . significant difference may appear if we consider the beyond - mean - field corrections , since for the @xmath181 case the scalar diquarks are not pseudo - goldstone bosons .
a calculation of the pion - pion scattering length in the @xmath338 channel can be performed within our theoretical framework .
the calculations of kaon condensation and kaon - kaon scattering length are also possible , but somewhat complicated due to the large mass difference between the light and strange quarks .
we are also interested in how the beyond - mean - field corrections modify the superfluid equations of state . as we learn from the knowledge of bec - bcs crossover in cold fermi gases ,
the superfluid equations of state can be strongly modified in the crossover regime @xcite , corresponding to the moderate baryon density in our case .
this issue is also important to the color - superconducting quark matter @xcite at moderate density , i.e. , for quark chemical potential around @xmath339mev where the pairing gap can be of order @xmath340mev@xmath341 .
the numerical works are in progress .
* acknowledgement : * the work is supported by the alexander von humboldt foundation . l. he would like to thank dr .
tomas brauner for valuable discussions and critical reading of the manuscript , and prof .
dirk rischke for encouragement during the work .
in this appendix , we briefly review the theory of molecular bose condensation in two - component fermi gases in the strong coupling limit .
while there exist many theoretical approaches @xcite to deal with this problem , we employ the field theoretical approach @xcite parallel to that used in this paper .
the lagrangian density of the system can be written as @xmath342 where @xmath343 denote the two - component ( nonrelativistic ) fermion fields with equal masses @xmath344 and chemical potentials @xmath345 . the gas is assumed to be dilute , and the coupling constant @xmath346 can be related to the s - wave fermion - fermion scattering length @xmath347 as @xmath348 where @xmath349 . in the dilute limit , we can take the limit @xmath350 in the final result . performing the hubbard - stratonovich transformation with the auxiliary boson field @xmath351 , and defining the nambu - gorkov representation @xmath352
, we can evaluate the partition function of the system as @xmath353[d\psi][d\phi^\dagger][d\phi]\exp{\left(-{\cal a}_{\text{eff}}\right)}$ ] , where @xmath354 and the inverse fermion propagator @xmath355 is given by @xmath356 then integrating out the fermionic degree of freedom , we get @xmath357[d\phi ] \exp{\left(-{\cal s}_{\text{eff}}\right)}$ ] where the bosonized effective action reads @xmath358=\int dx\frac{|\phi(x)|^2}{g}-\text{tr}\ln { \bf g}^{-1}(x , x^\prime).\end{aligned}\ ] ] first , we consider the mean - field theory where the auxiliary boson field @xmath359 is replaced by its expectation value @xmath360 . in the strong coupling limit @xmath361 , the fermion chemical potential @xmath345 approaches @xmath362 with @xmath363 being the molecular binding energy .
since the pairing gap @xmath364 , we can expand the effective action in powers of @xmath365 , which resulting in a ginzburg - landau free energy functional @xmath366=\int dx\bigg[\delta^\dagger(x)\left ( \kappa\frac{\partial}{\partial\tau}-\gamma\mbox{\boldmath{$\nabla$}}^2\right)\delta(x)\nonumber\\ & & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ + \ \ \alpha|\delta(x)|^2+\frac{1}{2}\beta|\delta(x)|^4\bigg].\end{aligned}\ ] ] the coefficients @xmath367 of the potential terms can be obtained from the mean - field thermodynamic potential @xmath368 $ ] which can be evaluated as @xmath369 where @xmath370 and @xmath371 . after a simple algebra ,
the coefficients @xmath372 and @xmath373 can be evaluated as @xmath374 from the expression of @xmath372 , we find that a quantum phase transition from vacuum to bose condensation takes place at @xmath375 .
thus , near the phase transition , @xmath372 can be simplified as @xmath376 where @xmath377 is the boson chemical potential .
further , setting @xmath378 , @xmath373 can be simplified as @xmath379 the coefficients @xmath380 of the kinetic terms can be obtained from the inverse boson propagator @xmath381 with @xmath130 . it can be evaluated as @xmath382 in the strong coupling limit , it can be well approximated as@xcite @xmath383 in summary , if we define the new condensate wave function @xmath384 by @xmath385 the ginzburg - landau free energy can be reduced to the gross - pitaevskii free energy of dilute bose gases , @xmath386=\int dx\bigg[\psiup^\dagger(x)\left ( \frac{\partial}{\partial\tau}-\frac{\mbox{\boldmath{$\nabla$}}^2}{2m_{\text b}}\right)\psiup(x)\nonumber\\ & & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\ \ \mu_{\text b}|\psiup(x)|^2+\frac{1}{2}\frac{4\pi a_{\text{bb}}}{m_{\text b}}|\psiup(x)|^4\bigg],\end{aligned}\ ] ] where @xmath387 is the boson mass and @xmath388 is the boson - boson scattering length . since @xmath361 ,
the interactions among the composite bosons are repulsive and weak .
to study the beyond - mean - field corrections , we consider the fluctuations around the mean field .
making the field shift @xmath389 , we can expand the effective action @xmath53 in powers of the fluctuations .
the zeroth order term @xmath390 is just the mean - field result , and the linear terms vanish automatically guaranteed by the saddle point condition for @xmath64 .
the quadratic terms , corresponding to gaussian fluctuations , can be evaluated as @xmath391 where the inverse boson propagator @xmath117 is given by @xmath392 and @xmath393 here the fermion green function @xmath88 is defined as @xmath394 $ ] and the bcs distribution functions @xmath395 and @xmath396 are used . in the strong coupling limit where
@xmath397 , the matrix elements of @xmath117 can be analytically evaluated .
we have @xcite @xmath398 where @xmath399 and @xmath400 is the minimum of the gross - pitaevskii free energy . together with the mean - field result for the boson density @xmath401
, we can show that the goldstone mode takes a dispersion relation given by @xmath402 which is just the bogoliubov excitation in a dilute bose condensate .
to evaluate the thermodynamic potential beyond the mean field , we express the partition function in the gaussian approximation as @xmath403[d\phi]\exp\big(-{\cal s}_{\text{eff}}^{(2)}\big).\end{aligned}\ ] ] integrating out the gaussian fluctuations , the total thermodynamic potential can be expressed as @xmath404 where the contribution from the gaussian fluctuations can be evaluated as @xcite @xmath405e^{i\nu_m0^+}.\end{aligned}\ ] ] near the quantum phase transition point @xmath406 , we can expand @xmath198 in powers of @xmath65 . because of the properties of second order phase transition , the terms of order @xmath266 vanish .
to leading order , the result is @xcite @xmath407 where the numerical factor @xmath408 .
from the gross - pitaevskii free energy , we find that the pressure @xmath278 in the mean - field approximation can be expressed as @xmath409 thus , to leading order , the beyond - mean - field corrections renormalize the boson - boson scattering length @xmath410 .
the new renormalized scattering length reads @xmath411 notice that this result is quite close to the exact result for the four body problem of @xmath412 @xcite .
this means the quantum fluctuations are almost correctly included in the present theoretical approach .
further , going beyond the leading order we find that we can fit @xmath198 to the functional form @xcite @xmath413 where @xmath414 and the dimensionless factors @xmath415 can be numerically determined .
solving for the molecular chemical potential @xmath416 one obtains @xmath417,\end{aligned}\ ] ] with the coefficient @xmath418@xcite which is @xmath419 smaller than the lee - huang - yang result @xmath420 @xcite .
in this appendix , we evaluate the explicit forms of the one - loop susceptibilities @xmath119 ( @xmath421 ) and @xmath136 . at arbitrary temperature ,
their expressions are rather huge .
however , at @xmath127 , they can be written in rather compact forms . for convenience ,
we define @xmath422 in this appendix . _
( i ) diquark sector .
_ first , the polarization functions @xmath423 and @xmath424 can be evaluated as @xmath425,\nonumber\\ \pi_{12}(q)&=&n_cn_f\sum_{\bf k}\bigg[\left(\frac{u_{\bf k}^-v_{\bf k}^-u_{\bf p}^-v_{\bf p}^-}{i\nu_m+e_{\bf k}^-+e_{\bf p}^-}-\frac{u_{\bf k}^-v_{\bf k}^-u_{\bf p}^-v_{\bf p}^-}{i\nu_m - e_{\bf k}^--e_{\bf p}^-}+\frac{u_{\bf k}^+v_{\bf k}^+u_{\bf p}^+v_{\bf p}^+}{i\nu_m+e_{\bf k}^++e_{\bf p}^+}-\frac{u_{\bf k}^+v_{\bf k}^+u_{\bf p}^+v_{\bf p}^+}{i\nu_m - e_{\bf k}^+-e_{\bf p}^+}\right){\cal t}_+ \nonumber\\ & & \ \ \ \ \ \ \ \ \ \ \ \ \ \ + \left(\frac{u_{\bf k}^-v_{\bf k}^-u_{\bf p}^+v_{\bf p}^+}{i\nu_m+e_{\bf k}^-+e_{\bf p}^+}-\frac{u_{\bf k}^-v_{\bf k}^-u_{\bf p}^+v_{\bf p}^+}{i\nu_m - e_{\bf k}^--e_{\bf p}^+}+\frac{u_{\bf k}^+v_{\bf k}^+u_{\bf p}^-v_{\bf p}^-}{i\nu_m+e_{\bf k}^++e_{\bf p}^-}-\frac{u_{\bf k}^+v_{\bf k}^+u_{\bf p}^-v_{\bf p}^-}{i\nu_m - e_{\bf k}^+-e_{\bf p}^-}\right){\cal t}_-\bigg]e^{2i\theta},\end{aligned}\ ] ] where @xmath426 . here
@xmath427 are factors arising from the trace in spin space , @xmath428 and @xmath429 are the bcs distribution functions defined as @xmath430 at @xmath431 , we find that @xmath432.\ ] ] thus , near the quantum phase transition point , we have @xmath433 . on the other hand , a simple algebra shows that @xmath434 therefore , the mean - field gap equation for @xmath64 ensures the goldstone s theorem in the superfluid phase . _
( ii ) diquark - sigma mixing terms . _
the term @xmath131 standing for the mixing between the sigma meson and the diquarks reads @xmath435e^{i\theta},\end{aligned}\ ] ] where the factors @xmath436 are defined as @xmath437 one can easily find that @xmath438 , thus it vanishes when @xmath64 or @xmath83 approaches zero . at @xmath431
, we have @xmath439.\ ] ] thus the quantity @xmath440 defined in ( [ bexp ] ) can be evaluated as @xmath441 _ ( iii ) sigma meson and pions . _
the polarization function @xmath442 which stands for the sigma meson can be evaluated as @xmath443,\end{aligned}\ ] ] where the factors @xmath444 are defined as @xmath445 at @xmath431 and for @xmath130 , we find that @xmath446 finally , the polarization function @xmath136 for pions can be obtained by replacing @xmath447 .
thus , when @xmath448 , the sigma meson and pions become degenerate and chiral symmetry is restored .
in this appendix , we derive the expression of the taylor expansion of @xmath198 in terms of @xmath449 . as we have shown in sec .
[ s5 ] , the leading - order term should be @xmath272 .
thus , we need to evaluate the numerical factor @xmath274 .
a key problem here is that the effective quark mass @xmath83 and the chemical potential @xmath0 are both functions of @xmath65 determined at the mean - field level . first , we expand the matrix elements of @xmath117 and @xmath134 in terms of @xmath450 . any of these elements denoted by @xmath451 is a function of @xmath0 , @xmath83 and @xmath450 .
our method of expansion is as follows .
we firstly expand @xmath452 in terms of @xmath450 formally with @xmath0 and @xmath83 being fixed parameters , i.e. , @xmath453 where @xmath454 and the expansion coefficients are defined as @xmath455 we then expand the coefficients @xmath456 ( @xmath457 ) at @xmath458 , using the fact that @xmath459 doing this we formally obtain @xmath460 finally , up to order @xmath461 , we have meanwhile , the thermodynamic potential @xmath198 can be expressed as @xmath466+\ln\det{\bf n}(q)\right\}e^{i\nu_m 0^+}-\frac{1}{2}\sum_q\left[2\ln{\cal d}_{\text d}^{*-1}(q)+\ln{\cal d}_\sigma^{*-1}(q)+3{\cal d}_\pi^{*-1}(q)\right]e^{i\nu_m 0^+}.\end{aligned}\ ] ] using the expansion ( [ c6 ] ) , we find that the factor @xmath274 is given by @xmath467 ^ 2-\frac{y_3(q)}{{\cal d}_{\pi}^{*-1}(q)}\right\}.\end{aligned}\ ] ] it is obvious that @xmath468 and @xmath469 where @xmath470 and @xmath471 are defined in ( [ bexp ] ) . on the other hand , since @xmath472 to leading order of @xmath473 , we can set @xmath474 in all equations and identify @xmath475 defined in ( [ bexp ] ) . | qcd - like theories possess a positively definite fermion determinant at finite baryon chemical potential @xmath0 and the lattice simulation can be successfully performed .
while the chiral perturbation theories are sufficient to describe the bose condensate at low density , to describe the crossover from bose - einstein condensation ( bec ) to bcs superfluidity at moderate density we should use some fermionic effective model of qcd , such as the nambu jona - lasinio model . in this paper , using two - color two - flavor qcd as an example , we examine how the nambu jona - lasinio model describes the weakly interacting bose condensate at low density and the bec - bcs crossover at moderate density .
near the quantum phase transition point @xmath1 ( @xmath2 is the mass of pion / diquark multiplet ) , the ginzburg - landau free energy at the mean - field level can be reduced to the gross - pitaevskii free energy describing a weakly repulsive bose condensate with a diquark - diquark scattering length identical to that predicted by the chiral perturbation theories .
the goldstone mode recovers the bogoliubov excitation in weakly interacting bose condensates .
the results of in - medium chiral and diquark condensates predicted by chiral perturbation theories are analytically recovered .
the bec - bcs crossover and meson mott transition at moderate baryon chemical potential as well as the beyond - mean - field corrections are studied .
part of our results can also be applied to real qcd at finite baryon or isospin chemical potential . |
evolution of a degenerate c o stellar core with a mass close to the chandrasekhar limit ( @xmath14 for a carbon white dwarf , see bisnovatyi - kogan , 1989 ) leads to the development of thermal instability and to the explosion .
the critical density @xmath15 for the beginning of the explosion may vary from @xmath16 to @xmath17 .
it increases with reduction of @xmath18 the core mass growth - rate which for a single asymptotic giant branch ( agb ) star is defined by the paczyski uus relation ( paczyski , 1970 ) @xmath19 and can take values from @xmath20 to @xmath21 in an accreting white dwarf ( iben , 1982 ; hachisu et al . , 1996 ) . in the present paper
we assume @xmath22 which can be appropriate both for a single agb - star and for a component of a binary system .
the explosion calculations , some of which accounted for convection , were carried by various authors ( arnett , 1969 ; ivanova et al . , 1974 ; nomoto et al . , 1976
; see also the review by niemeyer and woosley ( 1997 ) ) .
ivanova et al.(1974 ) found that carbon was burning in the deflagration regime with excitation of pulsations , but convection at the supernova stage was not considered , and the initial temperature profile was obtained not from evolutionary calculations , but from an estimate of a convection contribution .
woosley ( 1997 ) has carried out a series of one - dimensional explosion calculations for accreting white dwarfs ( unfortunately , in this work the accretion rates are not specified ) with critical central densities from @xmath23 to @xmath24 and with nuclear reactions for 442 isotopes , but he considered only the adiabatic convection at the presupernova stage , while at the beginning of the thermal runaway ( when @xmath25 k ) he turned off the convection artificially ( because further maintenance of the adiabatic temperature gradient in his model led to the prompt detonation ) , and defined the burning front propagation velocity according to woosley and timmes ( 1992 ) with a provision for a fractal dimension of the front .
many authors ( e.g. iben , 1982 ) use the hydrostatic system of equations with adiabatic convection in the stellar core , when calculating the presupernova evolution ( with time - scale @xmath26 yr ) which is valid for this stage , but is too crude at the beginning of the thermal runaway .
we carried our calculations both on the presupernova stage and on the stage of explosive burning by the hydrodynamic scheme elaborated in papers by blinnikov and rudzsky ( 1984 ) , blinnikov and bartunov ( 1993 ) , and by blinnikov and dunina - barkovskaya ( 1993 , 1994 ) . in the latter two papers
this hydrodynamic scheme was used for the calculation of white dwarf evolution .
here we have included non - adiabatic time - dependent convection in well - known mixing length approximation : @xmath27 /(\partial e/\partial t)_\rho\ , , \label{tdot}\\ \partial x_{\rm c}/\partial t & = & - x_{\rm c}^2 r_{\rm cc } + ( \partial x_{\rm c}/\partial t)_{\rm conv}\ ; , \label{xcdot}\\ \partial u_{\rm c}/\partial t & = & 2(v_{\rm c}^2-u_{\rm c}^2)/l_{\rm mix}\ ; , \label{ucdot}\end{aligned}\ ] ] where @xmath28 is the mass fraction of @xmath29 , @xmath30 is the energy release during the carbon burning , @xmath31 stands for standard neutrino energy losses ( as approximated by schinder et al . , 1987 ; see also haft et al . , 1994 ) , @xmath32 is the energy liberation during the gravitational contraction , @xmath33 is the radiative energy flux , @xmath34 is the convective energy flux , @xmath35 is the time - dependent convective velocity ( see arnett 1969 ) , and @xmath36 is the stationary convective velocity in the mixing length approximation which is given by the formula ( see , e.g. , bisnovatyi - kogan , 1989 ) @xmath37 the convective energy flux @xmath34 and the value @xmath38 , which corresponds to the change of a carbon mass fraction @xmath28 by convection , were calculated by the formulas @xmath39 @xmath40 the arrangement and the sizes of the convective zones are defined by the schwarzschild criterion ( according to the aforecited formulas ( [ vconv ] ) and ( [ fconv ] ) ) . on the presupernova stage
we did not account for time - dependent convection and
the formulas ( [ fconv][dxcconv ] ) contained @xmath36 instead of @xmath41 .
a comparison of calculations with stationary and time - dependent convection on the supernova stage was performed in ( dunina - barkovskaya , imshennik , 2000 ) .
the mixing length in the @xmath42-th mass zone was determined by the relation @xmath43 where @xmath44 is the size of the convective zone containing @xmath42-th zone , and @xmath45 in our calculations was taken equal to unity .
thermodynamic functions for the electron - positron gas ( the pressure @xmath46 , the entropy @xmath47 etc . ) were calculated with help of nadyozhin s asymptotic realtions described by blinnikov et al .
( 1996 ) , and for the ionic gas with screening by yakovlev and shalybkov ( 1988 ) .
the initial state of the presupernova c o core should to be taken from reliable evolutionary calculations .
we begin our computations with the central density @xmath48 which , according to iben ( 1982 ) , corresponds to the central temperature about @xmath49 k. everywhere below we define , as a beginning of the runaway and , thereafter , as an end of the presupernova stage , the moment of time when the temperature in the central zone reaches the value of @xmath50 .
the initial carbon mass fraction @xmath51 ( constant throughout the c o core ) was taken the same as in iben s ( 1982 ) work . assuming the above - mentioned central parameters for the c o core and its total mass @xmath0 , we have computed the adiabatic configuration in hydrostatic equilibrium which was used as an initial condition .
it should be mentioned that its difference from iben s ( 1982 ) model seems insignificant . during the calculations of the evolution of this model with account for a constant mass growth - rate @xmath21 ( typical for accreting white dwarfs in binary systems ) ,
it is worth - while to set a non - zero outer boundary condition for the pressure . in our previous work ( dunina - barkovskaya , imshennik , 2000 ) , where the calculations led to relatively small changes of the total mass before the runaway
, we included all the accreting mass in the outer boundary condition .
this condition is easily derived in the approximation of a thin ( @xmath52 ) and light ( @xmath53 ) envelope .
really , by integrating the hydrostatic equilibrium equation @xmath54 over the envelope thickness we obtain for the pressure @xmath55 on the outer boundary of the zone with the lagrangean coordinate @xmath56 : @xmath57 where @xmath58 and @xmath59 during the hydrostatic evolution , i.e. the outer pressure @xmath55 from ( [ pex ] ) increases in proportion to the time @xmath60 which is measured from the beginning of the c o core evolution calculations . in this case
the full mass @xmath61 ( where @xmath62 is the mass of @xmath63 lagrangean zones contained in the hydrodynamic model ) grew up to @xmath64 , in accordance with evolution time @xmath65 yr ( dunina - barkovskaya , imshennik , 2000 ) . in the present work
, we have gradually , according to the accretion , added new lagrangean zones to the model ( cf .
woosley , 1997 ) .
the mass of the zones decreased geometrically from @xmath66 ( 151-th zone ) down to @xmath67 ( the last , 170-th zone ) .
it was found in this case that the equation ( [ pex ] ) is quite inaccurate ( evidently because @xmath68 is not very small ) . as a result ,
the evolution time on the presupernova stage increased nearly twice .
for the variants with non - zero @xmath8 the mass of the model reached @xmath69 after @xmath70 yr , and the number of the lagrangean zones became @xmath71 .
after that we have added no lagrangean zones , but simulated accretion by increasing the outer pressure . in the runaway point
( after @xmath72 yr since the beginning of calculations ) the outer pressure corresponded to the additional external mass from @xmath73 ( for @xmath74 ) to @xmath75 ( for @xmath76 ) . in the control variant with @xmath77 the temperature in the central zone increased to @xmath50 already after @xmath78 years , therefore the model mass could grow only up to @xmath79 , which corresponds to the number of zones @xmath80 .
let us consider the boundary condition ( [ pex ] ) during the explosion for the variant with @xmath81 when @xmath82 , and @xmath83 cm ( fig .
1 ) : @xmath84 where @xmath85din @xmath86 . this value is to be compared with the central pressure of the c o core which has the well - known lower estimate @xmath87 .
thus , the ratio of the outer boundary pressure to the central one , @xmath88
@xmath89 @xmath90 , is surely very small at the beginning of the runaway . in the subsequent runaway process it certainly changes and can increase by factor of a few , but nevertheless remains small enough which can be seen from the numerical results .
it is easy to estimate that the relative contribution of the inertial term into the derived boundary condition ( [ pb133 m ] ) with characteristic parameters of c o core pulsations @xmath91 will be small : no more than a few percent .
thus we can neglect it in ( [ pex ] ) even when the pulsations amplitude is maximal @xmath92 and @xmath93 by the end of deflagration burning ( see fig .
this boundary condition imposed by the mass accretion onto the c o core surface is in close agreement with the outer boundary condition assumed in the previous works ( ivanova et al . , 1974 ;
ivanova et al . , 1977a ) with the same dependence on an outer radius @xmath94 :
@xmath95 ( [ pb133 m ] ) .
let us remind that in the cited works the outer pressure also simulated the presence of the outer c o core envelope .
it was taken small enough , but about a factor of 2 higher than the value from ( [ pb133 m ] ) .
thus , the effect of the outer boundary was more pronounced there than in the present work , but less significant than in ( dunina - barkovskaya , imshennik , 2000 ) .
it should be stressed that the non - zero outer pressure @xmath96 violates the energy conservation law during the explosion ( see fig . 4 , 5 ) when the pulsations arise and the radius @xmath94 of the c o core changes .
naturally , accounting for the accreting matter in the outer boundary condition ( [ pex ] ) and ( [ pb133 m ] ) we imply its attachment to the evolving c o core without any further energy release , i.e. merely an adhesion .
in fact , this substance has , generally speaking , another chemical composition , and can go through the thermonuclear reactions with an energy liberation and even with partial ejection of the accreted material .
we simply neglect those processes . on the other hand ,
the role of the outer boundary condition ( [ pb133 m ] ) is very significant during the explosion , especially when the detonation wave forms near the surface ( see below ) .
in the present work we have systematically investigated the models numerically with different values of the mixing length parameter ( [ lmixdri ] ) equal to @xmath97 ; @xmath98 ; @xmath99 ; @xmath100 ; @xmath101 . from a physical standpoint , it is very difficult to choose a certain value @xmath8 from this wide range in the used approximate theory of non - adiabatic convection .
the energy release in the thermonuclear reaction of carbon burning @xmath5c@xmath6c @xmath102mg@xmath103 was taken , first , by a rather crude formula obtained by fowler and hoyle ( 1965 ) with electron screening factor by salpeter ( 1954 ) ( see also salpeter and van horn 1969 , and arnett 1969 ) , and , second , with a maximum possible energy liberation corresponding to the instantaneous combustion of all the carbon - oxygen mixture to @xmath7ni . the energy release accepted here was taken from our early work ( ivanova et al .
it is convenient for the further comparison of our previous results with those obtained in the current work .
such a simplification of a thermonuclear energy generation can not affect essentially the obtained results due to the comparatively small sensitivity of the ignition conditions for the c o mixture to all pre - exponential values in the expression for the thermonuclear burning rate . as a result of the presupernova evolution calculations we obtained that just prior to the runaway the density drops to a value which depends non - monotonically on the parameter @xmath8 and
is maximal when @xmath97 .
in general , the initial critical central density ( in units @xmath48 ) @xmath104 turned out to be close to the one accepted earlier by ivanova et al .
( 1974 ) and later by ivanova et al .
( 1977a ) ( @xmath105 ) , though it is somewhat lower .
let us remind that in 1970s the most popular viewpoint was that single intermediate mass stars evolve along the convergent pazcyski s ( 1970 ) track after the formation of the carbon ( or carbon - oxygen ) core .
the modern approach to the presupernova evolution involves the evolution in a close binary system with an accreting white dwarf . in this case , according to yungelson ( 1998 ) , all the stars in the initial main sequence mass range @xmath106 may become carbon - oxygen white dwarfs in close binary systems .
this range does not differ significantly from the similar range for single stars @xmath107 considered previously ( pazcyski , 1970 ) . , the central density @xmath108 , the convective core mass @xmath109 , and the radius @xmath110 of the @xmath63-th ( last ) lagrangean zone upon time for @xmath111 after the convective core formation .
] some quantities characterizing the c o stellar core before the runaway are represented in fig . 1 : the convective core mass @xmath109 , the central temperature @xmath112 , the central density @xmath108 , and the radius of the c o core @xmath110 over the evolution time from @xmath113 yr ( the beginning of the convective core formation ) to @xmath114 yr for @xmath81 . small jumps in the radius of the last ( @xmath63-th )
lagrangean zone shown on fig .
2 clearly indicate the procedure of adding of new lagrangean zones described above .
the phase of the slow expansion before the runaway ( see above ) is almost unnoticeable on this graph because its duration ( @xmath115 yr ) is small in comparison to the overall evolution time ( during which the star , as a whole , was contracting very slowly ) and practically does not depend on @xmath8 . ) for various values of the parameter @xmath8 . ] in fig . 2 ,
the initial profiles of temperature depending on the mass coordinate are represented at the moment of the beginning of explosion ( see above ) .
it can be seen that the temperature in the convective variants is appreciably higher than that in the variant without convection .
even for @xmath116 the temperature exceeds this value almost by a factor of 2 in the central part of the convective core .
let us notice that in this moment there is an active combustion only in the first mass zone . on fig .
2 we compare the initial temperature profiles with data from ivanova et al .
( 1977b ) . in the latter work the explosion developed for the two ( upper ) @xmath117-profiles .
therefore , it is not surprising that the runaway led to the total disruption of the star for all @xmath118 in the present calculation ( see below ) . for the variants with @xmath119 and with @xmath116 which are the boundary values for the deflagration regime with pulsations ( see below ) ,
the adiabatic temperature profiles are also shown , with the entropy equal to the entropy of the 2nd mass zone for each variant .
it can be seen that our profiles lie quite below the corresponding adiabatic profiles , and they have another shape for @xmath120 , which of course can affect the development of the runaway without the regime of spontaneous burning ( blinnikov , khokhlov , 1986 ) . and of the full energy of the c o core @xmath121 for @xmath97 and @xmath10 ( prompt detonation ) .
] it is interesting that in the second mass zone ( with @xmath122 ) the smooth temperature profile for the maximum @xmath97 touches the third ( lower ) steep profile for which the explosion in ( ivanova et al . , 1977b ) failed .
but the initial profile for the variant with the minimal @xmath123 which in fact `` hardly '' exploded ( see fig .
5 ) already lies noticeably below the third profile from ( ivanova et al .
, 1977b ) for @xmath124 ( in the first 4 mass zones ) .
we can conclude that there is a qualitative agreement between our present results and the calculations in the cited work .
the development of the runaway was different for @xmath125 ( fig . 3 ) and for @xmath126 ( fig .
4 ) . on the said graphs
the time dependence of the central temperature @xmath112 and of the full energy of the c o core @xmath121 on time is shown .
the latter eventually becomes equal to the value @xmath127 corresponding to the full incineration into @xmath7ni for all the variants regardless of the value @xmath8 .
the value of @xmath121 at the moment of the runaway beginning ( @xmath128 ) is also practically the same for all this variants . and of the full energy of the c o core @xmath121 for @xmath129 and @xmath130 ( pulsational deflagration with delayed detonation ) .
] and of the full energy of the c o core @xmath121 for @xmath131 ( pulsational deflagration with delayed detonation ) . ]
( pulsational deflagration ) . ]
( prompt detonation ) . ]
( before the detonation ) .
convective zones are shown by labels and numbers , e.g. 2 - 1 means the first convective zone for the second moment ( t=0.6222 s ) . ]
the basic result of our numerical calculations is that the pulsational deflagration regime eventuates when the mixing length is small enough , namely @xmath126 .
it was found that for @xmath132 the prompt detonation occurs ( see fig .
hereby we obtained a reasonably accurate threshold for the transition from the detonation to the pulsational deflagration . on fig .
4 there is 10 pulsations for the variant with @xmath133 and 14 pulsations for the variant with @xmath134 . in both cases
the burning ends with a powerful energy release in the last pulsation . in this case
the abrupt change of the burning regime occurs . in the previous pulsations
there was a deflagration regime , and in the last pulsation the all remaining fuel ( about 90% ) is burning in the regime of detonation .
the value @xmath112 reaches its maximal value @xmath135 a bit later , immediately after the collision of the detonation front propagating inward from the surface and deflagration front , during the afterburning in the central zone ) .
thus , the hybrid regime of the thermonuclear explosion of a carbon - oxygen white dwarf takes place , which is most promising for the explosive nucleosynthesis ( niemeyer and woosley , 1997 ) . on the other hand ,
3 demonstrates the burning without pulsations in an ordinary detonation regime starting from the center ( for @xmath136 ) . in fig . 5 ,
the time dependence of the central temperature and of the total energy is shown for the computation with @xmath137 . here
the total number of pulsations is appreciably larger than on the previous graph , but only 14 of them were accompanied by the propagation of deflagration into a neighboring mass zone , which can be well seen by very narrow specific maxima of the temperature @xmath138 .
let us notice that these narrow maxima correspond ( due to the energy generation ) to the small jumps of the value @xmath121 . in the variant with @xmath13 one to three mass zones were burning during each pulsation ( except for 5th and 7th ) ,
so almost all the pulsations had the said temperature peaks ( fig . 4 ) .
the deflagration regime is distinctly characterized by the profiles of density which are represented versus the eulerian radius @xmath139 in fig . 6 for @xmath13 .
the density drops by a factor of @xmath140 at the deflagration front .
depending upon the pulsation phase the front either moves inward ( the contraction phase ) or outward ( the expansion phase ) during the overall propagation ( by mass zones ) .
the pulsations can be seen more clearly by the behaviour of the outer layers of the star , in the still unburnt matter .
the amplitude of the outer radius pulsations grows in time and attains the value of @xmath141 ( this can be seen well due to the logarithmic scale for density in fig . 6 ) .
the density profiles @xmath142 for stronger convection are given in fig . 7 for @xmath143 .
the profiles of the temperature @xmath144 are plotted on this graph in a small scale , they show neatly the position of the burning front as a jump of @xmath117 .
it can be seen that at the burning front the density of matter distinctly increases displaying the detonation burning regime .
the supersonic character of detonation is also revealed by the quiescence of stellar layers outside the burning front .
the convective processes are most neatly characterized by the profiles of the specific entropy which are shown in fig . 8 for the case @xmath13 versus the mass coordinate @xmath145 ( in logarithmic scale ) .
the convective zones which are defined by the known schwarzschild criterion , i.e. correspond to the regions with the negative entropy gradient , are marked on these profiles for some typical moments of time .
this figure shows that the primary convective core during the runaway is being separated into some convective zones , the largest of which includes one of two mass zones behind the burning front and the rest of the convective core before the front .
the separation process is of nonstationary character and indicates the importance of the time - dependent convection .
the entropy profiles are given in fig .
8 before the formation of the detonation wave propagating towards the center ( see fig . 4 ) . after the formation inception of such a wave , naturally , in a very short time @xmath146 s , the entropy behind the detonation front rises to the values which are typical also for the deflagration , @xmath147 , i.e. by factor of 4 - 6 more than an initial entropy ( @xmath148 , with a positive gradient . only in outer layers
( with @xmath149 ) the negative entropy gradient persists , and , therefore , the convection occurs , partly preventing the detonation process described here .
the important role of the convection intensity characterized by a parameter @xmath8 attracts our special attention .
it is clear that the deflagration burning regime with pulsations ( of course , the deflagration is not necessarily accompanied by pulsations , but for the latter the subsonic deflagration is the necessary condition ) takes place in the whole range of the @xmath8 values , @xmath150 .
it is essential that @xmath151 is above zero , because for @xmath77 in our calculation there was no hydrodynamic explosion at all , although the first mass zone was burning .
we note that for higher critical densities @xmath152 which is the case for lower accretion rate ( see above ) , the value of @xmath151 can drop practically down to zero ( zmitrenko et al . , 1978 ) .
however , for high densities @xmath15 the account for neutrino energy losses together with the neutronization kinetics is required ( ivanova et al .
, 1977a , c ) . but in the present case ( @xmath153 ) the role of these complicated processes is not important , which was shown by direct comparison of the calculations of ivanova et al .
( 1974 ) with those of ivanova et al .
( 1977b ) at the close value of central density @xmath154 .
thus , the account for convection processes can prevent the pulsations development during the deflagration burning when the parameter @xmath8 exceed some critical value @xmath155 . in the present work it is found that @xmath156 .
during the delayed detonation ( up to the beginning of last pulsation with the detonation initiation ) . ] ( pulsational deflagration up to the detonation initiation ) . ] . ] in essence , the emergence of the detonation , after a long period of the deflagration burning accompanied by the development of the global pulsations of the c o core , can be termed a `` delayed detonation '' ( niemeyer and woosley , 1997 ) : from the moment of the central ignition a long time ( tens of seconds ) passes , i.e. many thermodynamic time - scales of the order of the pulsation period , 2 - 3 s. an analysis of the results has shown that the detonation burning front emerges in a mass coordinate about @xmath0 , more specifiacally , in the 150th mass zone when the total number of zones equals 170 .
this coordinate is virtually independent of the parameter @xmath8 . in our model in the 150th zone the mass grid changed from uniform to geometrically decreasing . in the work ( dunina - barkovskaya , imshennik , 2000 ) the 127th mass zone in which the detonation begins had the same feature
, so we have to investigate the influence of a mass grid nonuniformity on a detonation initiation in our future work . in the present section
we shall try to give a physical justification of the entropy growth that leads to the detonation .
after some first pulsations , i.e. long before the generation of the burning front , the entropy of the outer mass zones begins to increase ( fig . 9 ) .
at the moment @xmath157 s ( the beginning of the last pulsation ) two local entropy maximums exist , the outermost one ( @xmath158 ) and the innermost one ( @xmath159 ) .
though the entropy in the innermost maximum is less than in the outermost one , the temperature in the innermost maximum during the last pulsation is higher ( due to the higher density ) . just at this maximum
the detonation burning front is born ( see fig .
we can see the relatively high growth of entropy between the close moments @xmath160 s and @xmath161 s , between the maximum and the minimum of the second ( ! ) pulsation ( see fig .
4 ) , which is less by amplitude than the subsequent pulsations .
it can be seen on the graphs for density ( fig .
10 ) that the moments close to the maxima of pulsations are accompanied by a subsequent emergence of the compression waves on the stellar surface . following ivanova et al .
( 1982 ) , one can estimate the entropy growth @xmath162 treating these compression waves as weak shocks with a pressure jump @xmath163 .
the one gets the well - known entropy jump ( landau and lifshitz , 1954 ) : @xmath164 this estimate is easy to apply in the case under consideration because we can account only for the pressure of the completely degenerate ultrarelativistic electron gas @xmath165 , with @xmath166@xmath167g@xmath168s@xmath169 , in calculation of the second adiabatic derivative in ( [ ds ] ) , whence it follows that @xmath170 such an approximation is possible because in the zone under consideration with @xmath171 we can take as an estimate for the values @xmath172 and @xmath173 in front of the involved weak shock their values at the moment @xmath160 s , namely , @xmath174 , and @xmath175k . after a simple rearrangement we get the final expression for the specific entropy growth in ( [ ds ] ) ( @xmath47 is in units @xmath176 , see fig .
9 ) : @xmath177 ^ 3.\label{ds8}\end{aligned}\ ] ] in the expression ( [ ds8 ] ) it is necessary to specify also the amplitude of the weak shock wave . from fig .
10 we can get @xmath178 ( for the moment @xmath161 s )
. then we finally get from ( [ ds8 ] ) : @xmath179 , whereas in fig .
9 the entropy growth between these two moments is @xmath180 .
thereby we can reasonably explain the entropy growth obtained in calculations as a result of a weak shock dissipation which occurs in the region of a rather abrupt density decrease .
closer to the surface the entropy growth subsides , according to ( [ ds8 ] ) , because of the density lowering towards the star edge , and inward it subsides too , probably due to the wave amplitude reduction ( see also fig . 10 ) . nevertheless , the local entropy maximum ( fig .
9 ) does not disappear with time , but persists and increases furthermore by a factor of @xmath181 before the moment of the detonation generation .
it reaches the value of @xmath182 .
let us notice that unfortunately this can not be seen on fig .
8 , because the outer stellar layer are shown there in the logarithmic scale by mass . in fig .
11 , we represent the temperature profiles from the moment of the burning front initiation ( @xmath183 ) to its `` collision '' with the deflagration front at a radius @xmath184 cm ( which corresponds to the mass from the center @xmath185 ) at the moment @xmath186 s. on this graph the beginning of the expansion of matter behind the detonation wave can be well seen : the outer radius rises from @xmath187 cm to @xmath188 cm at the temperature @xmath189 .
when the detonation wave propagates in the direction of increasing density , the temperature rises to @xmath190 , partly due to the heat capacity decline .
the compression on the front is typically modest : only by a factor of 1.5 ( see fig .
the pressure jump is not high too : about a factor of 2 till the end .
it is remarkable that the generation of the detonation front really occurs at the innermost entropy maximum ( see fig .
9 ) , which corresponds in fig .
11 to the innermost temperature maximum at the moment @xmath157 s. a fair question arises : whether such a detonation front is stable with respect to galloping instability considered recently by imshennik et al .
this question requires an additional analysis , but it is conceivable that the stability here is more natural , because the density in front of the detonation wave increases which counteracts the decoupling of the shock wave front from the burning zone .
there is one more argument in favor of the stability against the multidimensional disturbances , namely , the large enough width of the burning zone which can be compared with the c o core radius at the densities @xmath191 ( imshennik , khokhlov , 1984 ) .
this implies also that the burning of the c o mixture as a matter of fact does not fuse the elements to `` iron '' ( `` fe '' ) in the converging detonation wave , but is limited , for example , by such nuclides as si etc .
the problems of nucleosynthesis in the obtained regime of delayed detonation should be investigated in other works .
in the current study , first of all , we obtained the initial conditions , justified by evolution calculations , for the thermonuclear runaway in the carbon - oxygen stellar core with a mass close to the chandrasekhar limit due to the accretion prescribed by a constant value @xmath192 , which does not contradict to the current evolutionary studies of close binary systems . in our simulations
the runaway begins at the moment defined by the calculation itself ( and the equations do not change , except for the account for the time - dependent convection ) .
thus , we do not have to solve a complicated problem of the choice of the initial conditions , which largely affected the inception of the runaway . in our previous work ( ivanova et al .
, 1974 , 1977a , b , c ) the initial conditions ( i. e. the initial temperature profiles ) were defined , strictly speaking , without an adequate justification .
the present computations include a well - known approximate model of a nonadiabatic convection , namely , the mixing length theory with the only arbitrary parameter @xmath8 ( a ratio of the mixing length @xmath193 to the radial extent of the convective zone @xmath44 ) .
the value of this parameter could be , in principle , extracted from the multidimensional approaches developed currently ( see lisewski et al . , 2000 ) , i. e. the arbitrariness of the approximate theory could be removed . as a result of our computations
it was found that the delayed detonation regime takes place under quite a wide range of values @xmath194 .
the lower boundary of this interval is defined as an upper estimate . at higher values of a parameter @xmath195 in our calculations the ordinary detonation propagating from the center of the stellar c
o core was obtained , which is unlikely to take place because of the instability of the burning front etc .
it could be interesting to find out whether the specified interval intersects with the region of effective mixing length parameters justified by a multidimensional theory of turbulence .
but already with the parametric simulations of the flame propagation velocity we can impose some restrictions on the parameter @xmath8 . for example , we can notice that for @xmath196 during the burning of a few first mass zones this velocity was close to the laminar front velocity obtained by woosley and timmes ( 1992 ) . in the present work , due to an application of the new hydrodynamic code with a variable lagrangean ( difference ) grid and with account for an outer boundary pressure conditioned by accretion , we have obtained a scenario of the development of a delayed detonation propagating from the surface towards the center of the star .
the detonation is , in all probability , stable with respect to the galloping instability , and is conditioned by a preceding stage of the pulsational deflagration .
this may prove to be important because of the recently emerged scepticism ( lisewski et al .
, 2000 ) in regard to the scenario of the deflagration to detonation transition due to the destruction of the laminar burning front by turbulent eddies published earlier ( khokhlov et al . , 1997 ) .
arnett w.d .
// astrophys .
space sci .
, 1969 , v.5 , p.180 .
bisnovatyi - kogan g.s .
// physical problems of the theory of stellar evolution .
moscow : nauka , 1989.(in russian ) blinnikov s.i .
, bartunov o.s .
// astron .
astrophys . , 1993 ,
v.273 , p.106 .
blinnikov s.i .
, dunina - barkovskaya n.v.// astronomy reports , 1993 , v.37 , p.187 .
blinnikov s.i .
, dunina - barkovskaya n.v .
// mnras , 1994 , v.266 , p.289 .
blinnikov s.i .
, dunina - barkovskaya n.v .
, nadyozhin d.k . //
astrophys.j.suppl . , 1996 , v.106 , p.171
blinnikov s.i . ,
khokhlov a.m. // sov.astron.lett .
, 1986 , v.12 , p.131 .
blinnikov s.i .
, rudzsky m.a . //
1984 , v.10 , p.152 .
dunina - barkovskaya n.v .
, imshennik v.s.// proceedings of lebedev physical institute , 2000 , v.227 , p.32 .
fowler w. , hoyle f. //
nucleosynthesis in massive stars and supernovae .
chicago , university of chicago press .
hachisu i. , kato m. , nomoto k. // astrophys . j. lett . , 1996 , v.470 , l97 .
haft m. , raffelt g. , weiss a. //astrophys.j .
, 1994 , v.425 , p.222 .
iben i. // astrophys.j . , 1982 , v.253 , p.248 .
imshennik v.s . , kalyanova n.l . ,
koldoba a.v .
, chechetkin v.m .
// astronomy letters , 1999 , v.25 , p.206 .
imshennik v.s .
, khokhlov a.m. // sov.astrom.lett .
, 1984 , v.10 , p.262 .
ivanova l.n . ,
imshennik v.s .
, chechetkin v.m .
// astrophys .
space sci .
, 1974 , v.31 , p.497 .
ivanova l.n . , imshennik v.s . , chechetkin v.m . // sov.astron . , 1977a , v.21 , p.197 .
ivanova l.n . ,
imshennik v.s .
, chechetkin v.m . //
sov.astron . , 1977b , v.21 , p.374 .
ivanova l.n . ,
imshennik v.s .
, chechetkin v.m . //
sov.astron . , 1977c , v.21 , p.571 .
ivanova l.n .
, imshennik v.s .
, chechetkin v.m . //
sov.astron.lett . ,
v.8 , p.8 .
khokhlov a.m. , oran e.s .
, wheeler j.c.// astrophys.j . , 1997 , v.478 , p.678 .
landau l.d .
, lifshits e.m . //
mechanics of continua .
moscow : gos.izd-vo tekhniko-teor.lit.,1954 .
( in russian ) lisewski a.m. , hillebrandt w. , woosley s.e . //
astrophys.j . ,
2000 , v.538 , p.831 .
niemeyer j.c . , woosley s.e . //
astrophys.j . , 1997 ,
v.475 , p.740 .
nomoto k. , sugimoto d. , neo s. //
space sci .
, 1976 , v.39 , p.l37 . paczyski b. //
acta astronomica , 1970 , v.20 , p.47 .
salpeter e.e . //
australian j.phys . , 1954 ,
v.7 , p.373 .
salpeter e.e . , van horn h.m . //astrophys.j . , 1969 ,
v.155 , p.183 .
schinder p.j . ,
schramm d.n . ,
wiita p.j . ,
margolis s.h . ,
tubbs , d.l.// astrophys.j .
, 1987 , v.313 , p.531 .
woosley s.e . //
, 1997 , v.476 , p.801 .
woosley s.e .
, timmes f.x . //
, 1992 , v.396 , p.649 .
yakovlev d.g .
, shalybkov d.a .
// itogi nauki i tekh .
astronomia , 1988 , v.38 , p.191 .
( in russian ) yungelson l.r .
// contemporary problems of stellar evolution .
proceedings of the international conference in zvenigorod .
october 1315 1998 , p.79 .
zmitrenko n.v . ,
imshennik v.s . ,
khlopov m.yu .
, chechetkin v.m . //
i teor.fiz . , 1978 , v.48 , p.589 .
( in russian ) | a model of a carbon - oxygen ( c o ) presupernova core with an initial mass @xmath0 , an initial carbon mass fraction @xmath1 , and with an average mass growth - rate @xmath2 due to accretion in a binary system was evolved from initial density @xmath3 , and temperature @xmath4k through convective core formation and its subsequent expansion to the carbon runaway at the center . the only thermonuclear reaction contained in the equations of evolution and runaway was the carbon burning reaction @xmath5c@xmath6c with an energy release corresponding to the full transition of carbon and oxygen ( with the same rate as carbon ) into @xmath7ni . as a parameter we take @xmath8
a ratio of a mixing length to the size of the convective zone . in spite of the crude assumptions , we obtained a pattern of the runaway acceptable for the supernova theory with the strong dependence of its duration on @xmath8 . in the variants with large enough values of @xmath9 and
@xmath10 the fuel combustion occurred from the very beginning as a prompt detonation . in the range of @xmath11
the burning started as a deflagration with excitation of stellar pulsations with growing amplitude . eventually , the detonation set in , which was activated near the surface layers of the presupernova ( with @xmath12 ) and penetrated into the star down to the deflagration front .
excitation of model pulsations and formation of a detonation front are described in detail for the variant with @xmath13 .
_ keywords : _ supernovae and supernova remnants ; plasma astrophysics ; hydrodynamics and shock waves ; detonation and deflagration . |
in recent years a significant effort has been devoted to the understanding of glass - forming systems .
recent theoretical and numerical results clearly show that the slowing down of the dynamics near the glass transition is strongly connected to the potential energy landscape geometry .
the trajectory of the representative point in the configuration space can be viewed as a path in a multidimensional potential energy surface @xcite .
the dynamics is therefore strongly influenced by the topography of the potential energy landscape : local minima , barriers heights , basins of attraction an other topological properties all influence the dynamics .
the potential energy surface of a super - cooled liquid contains a large number of local minima , called _ inherent structures _ ( is ) by stillinger @xcite . all states that under local energy minimization will flow into
the same is define the _ basin _ of the is ( valley ) . with this pictures in mind
the time evolution of the system can be seen as the result of two different processes : thermal relaxation into basins ( _ intra - basin _ motion ) and thermally activated potential energy barrier crossing between different basins ( _ inter - basin _ motion ) . when the temperature is lowered down to the order of the critical mode coupling theory ( mct ) temperature @xmath1 the inter - basin motion slows down and the relaxation dynamics
is dominated by the slow thermally activated crossing of potential energy barriers @xcite .
if the temperature is further reduced the relaxation time eventually becomes of the same order of the observation time and the system falls out of equilibrium since there is not enough time to cross barriers and equilibrate .
this define the `` experimental''glass transition temperature @xmath2 . the regime between @xmath1 and @xmath2 can not be described by the mct since it neglects activated processes responsible for barrier crossing . in mct
the relaxation time diverges at @xmath1 , leading to @xmath3 , and the dynamics remains confined into a single basin forever .
the essential features of mct for glass - forming systems are also common to the high temperature phase of some fully connected spin glass models @xcite , the most well known being the spherical @xmath0-spin spin glass model @xcite
. we shall call these models _ mean - field @xmath0-spin - like _ glass models . as a consequence at the critical temperature @xmath1 , called @xmath4 in @xmath0-spin language
, an ergodic to non - ergodic transition takes place .
below this temperature the system is dynamically confined to a metastable state ( a basin ) @xcite since relaxation to true equilibrium can only take place via activated processes , absent in mean - field models .
for these systems , nevertheless , it is known that the true equilibrium transition to a low temperature phase occurs below @xmath4 at the static critical temperature @xmath5 , also denoted by @xmath6 @xcite .
this is the analogous of the kauzmann temperature @xmath7 for liquids .
the glass transition temperature @xmath2 of real systems sits somewhere in between @xmath5 and @xmath4 .
this transition , obviously , can not be reached even on infinite time in mean - field models . despite these difficulties mean - field models , having the clear advantage of being analytically tractable ,
have been largely used to study the properties of fragile glassy systems , especially between the dynamical temperature @xmath4 and the static temperature @xmath5 . the picture that emerges is however not complete since activated process can not be captured by mean - field models .
therefore the relevance of mean - field results for real systems can not be considered completely stated .
only recently activated processes in mean - field - like models have been invesigated in extended numerical investigation of _ finite - size _ fully - connected @xmath0-spin - like models @xcite . comparing the results with the observed behavior of super - cooled liquids near @xmath1
we can conclude that , once activated process are allowed , mean - field @xmath0-spin - like models are highly valuable for a deep understanding of the glass transition in real systems .
we report the main results obtained for the ising - spin random orthogonal model ( rom ) @xcite , defined by the hamiltonian @xcite , @xmath8 where @xmath9 are @xmath10 ising spin variables , and @xmath11 is a @xmath12 random symmetric orthogonal matrix with @xmath13 . for @xmath14
this model has the same thermodynamic properties of the @xmath0-spin model : a dynamical transition at @xmath15 , with threshold energy per spin @xmath16 , and a static transition at @xmath17 , with critical energy per spin @xmath18 @xcite .
the free energy analysis ( tap ) @xcite reveals that the phase space is composed by an exponentially large ( in @xmath10 ) number of different basins , separated by infinitely large ( for @xmath14 ) barriers .
each basin is unambiguously labelled by the value of the energy density @xmath19 of the local minimum contained within it , i.e. the is of the system . in this picture
the dynamical transition is associated with is having the largest basin of attraction for @xmath14 , while the static transition with is with the lowest accessible free energy ( vanishing configurational entropy ) @xcite . in the mean - field limit ,
the allowed values of @xmath19 are between @xmath20 and @xmath21 .
solutions with @xmath19 larger than @xmath21 are unstable ( saddles ) , while solutions with @xmath19 smaller than @xmath20 have negligible statistical weight .
moreover in the @xmath14 limit is with @xmath22 attract most ( exponentially in @xmath10 ) of the states and dominate the behavior of the system .
other is are irrelevant for @xmath14 . for finite @xmath10
the scenario is different since not only the basins of is with @xmath23 acquire statistical weight , but it may happen that solutions with @xmath24 and few negative directions ( saddles with few downhill directions ) become stable , simply because there are not enough degrees of freedom to hit them . to get more insight the is - structure of finite systems we follow stillinger and weber @xcite and decompose the partition sum into a sum over basins of different is and a sum within each basin .
collecting all is with the same energy @xmath19 , denoting with @xmath25\,de$ ] the number of is with energy between @xmath19 and @xmath26 , and shifting the energy of each basin with that of the associated is , the partition sum can be rewritten as @xcite @xmath27\ ] ] where @xmath28 can be seen as the free energy density of the system when confined in one of the basin associated with is of energy @xmath19 .
the function @xmath29 is the _ configurational entropy density _ also called _
complexity_. from the partition function we can compute the average internal energy density @xmath30 @xmath31 .
the first term is the average energy of the is relevant for the thermodynamics at temperature @xmath32 , while the second is the contribution from fluctuations inside the associated basins . in the limit
@xmath14 only is with @xmath22 contribute and @xmath33 for any @xmath34 . for finite @xmath10 , and @xmath32 not too close to @xmath4 ,
the thermodynamics is dominated by is with @xmath24 and @xmath35 @xcite .
this is more evident from the ( equilibrium ) probability distribution of @xmath19 since it is centered about @xmath36 indicating that is with @xmath37 have the largest basins .
this scenario has been also observed in real glass - forming systems@xcite . from the knowledge of is - energy distribution
we can reconstruct the complexity @xmath29 since from eq .
( [ eq : part ] ) the probability that an equilibrium configuration at temperature @xmath38 lies in a basin associated with is of energy between @xmath19 and @xmath26 is : @xmath39 / z_n(t ) .
$ ] in the temperature range where this applies , the curves @xmath40 are equal , except for a temperature dependent factor @xmath41 , to @xmath42 . if the @xmath19-dependence of @xmath28 can be neglected , then it is possible to superimpose the curves for different temperatures , see fig .
[ fig : f2 ] ( a ) .
the data collapse is rather good for @xmath43 . above the curves
can not be superimposed anymore indicating that the @xmath19-dependence of @xmath28 can not be neglected . in liquid
this is called the anharmonic threshold @xcite .
direct consequence of @xmath44 for @xmath45 is that in this range the partition function can be written as the product of an intra - basin contribution [ @xmath46 ) ] and of a configurational contribution which depends only on the is energy densities distribution .
the system can then be considered as composed by two independent subsystems : the intra - basin subsystem describing the equilibrium when confined within basins , and the is subsystem describing equilibrium via activated processes between different basins .
as the temperature is lowered and/or @xmath10 increased the two processes get more separated in time and the separation becomes more and more accurate . a scenario typical of super - cooled liquids near the mct transition @xcite .
more informations on the is structure can be obtained from non - equilibrium relaxation processes . to study the non - equilibrium dynamics we quench at time zero the system from an initial equilibrium configuration at temperature @xmath47 to a final temperature @xmath48 and study the evolution of the average is energy per spin @xmath49 as function of time ,
see fig.[fig : e - aging ] ( b ) .
two different relaxation processes are clear seen . a first regime independent of @xmath50 , and a second regime independent of both @xmath51 and @xmath50 .
the final temperature @xmath50 controls the cross - over between the two regimes .
a similar behavior has been observed in molecular dynamics simulations of super - cooled liquids @xcite .
the two regimes are associated with different relaxation processes . in the first part
the system has enough energy and relaxation is mainly due to _ path search _ out of basins through saddles of energy lower than @xmath52 .
this part depends only on the initial equilibrium temperature @xmath51 since it sets the initial phase space region .
different @xmath51 leads to different power law .
in particular relaxation must slow down as @xmath51 decreases since we expect that lower states are surrounded by higher barriers , in agreement with numerical data @xcite . during this process
the system explores deeper and deeper valleys ( basins ) while decreasing its energy .
the process stops when all barrier heights become of @xmath53 . from now on the relaxation proceeds only via activated process .
a first consequence is that lower the final temperature @xmath50 shorter the first relaxation , in agreement with our findings [ see figures [ fig : e - aging ] ] .
the analysis of the distance between the instantaneous system state and the corresponding is , counting the number of single spin flip needed to reach the is , reveals that for all times the systems stays in configurations few spin flips away from an is .
a similar study starting from equilibrium configurations at temperature @xmath54 evaluated comparing panels ( a ) and ( b ) of figure [ fig : e - aging ] @xcite leads to similar numbers .
we then conclude that during relaxation the aging system explores the same type of minima ( and basins ) visited in equilibrium at temperature @xmath55 .
direct consequence is that once the system has reached the activated regime there can not be memory of the initial @xmath51 , and all curves with different @xmath51 but same @xmath50 should collapse for large time @xcite .
to summarize , we have shown that _
finite - size _ mean - field @xmath0-spin - like models are good candidates for studying the glass transition .
the key point is that near the glass transition the thermodynamics of the systems is dominated by the is distributions , therefore all systems with similar is distributions should have similar behavior .
finite - size mean - field @xmath0-spin - like models have the double advantage of being analytically tractable for @xmath14 and easily simulated numerically for finite @xmath10 , offering good models to analyze the glass transition . | we analyze the properties of the energy landscape of _ finite - size _ fully connected @xmath0-spin - like models whose high temperature phase is described , in the thermodynamic limit , by the schematic mode coupling theory of super - cooled liquids .
we show that _ finite - size _ fully connected @xmath0-spin - like models , where activated processes are possible , do exhibit properties typical of real super - cooled liquid when both are near the critical glass transition .
our results support the conclusion that fully - connected @xmath0-spin - like models are the natural statistical mechanical models for studying the glass transition in super - cooled liquids . and glass transition , spin - glass , random models pacs : 64.70.pf , 75.10.nr , 61.20.gy , 82.20.wt |
[ sec : intro ] many interesting phenomena in the physical sciences and in applications are characterized by their high dimensionality and the presence of many different spatial and temporal scales .
standard examples include atmosphere and ocean sciences @xcite , molecular dynamics @xcite and materials science @xcite .
the mathematical description of phenomena of this type quite often leads to infinite dimensional multiscale systems that are described by nonlinear evolution partial differential equations ( pdes ) with multiple scales .
often physical systems are also subject to noise .
this noise might be either due to thermal fluctuations @xcite , noise in some control parameter @xcite , coarse - graining of a high - dimensional deterministic system with random initial conditions @xcite , or the stochastic parameterization of small scales @xcite .
high dimensional multiscale dynamical systems that are subject to noise can be modeled accurately using stochastic partial differential equations ( spdes ) with a multiscale structure .
there are very few instances where spdes with multiple scales can be treated analytically .
the goal of this paper is to develop numerical methods for solving accurately and efficiently multiscale spdes .
several numerical methods for spdes have been developed and analyzed in recent years , e.g. @xcite , based on a finite difference scheme in both space and time .
it is well known that explicit time discretization via standard methods ( e.g. , as the euler - maruyama method ) leads to a time - step restriction due to the stiffness originating from the discretisation of the diffusion operator ( e.g. the cfl condition @xmath0 , where @xmath1 and @xmath2 are the time and space discretization , respectively ) .
the situation is even worse for spdes with multiple scales ( e.g. of the form and below ) as in this case the courant - friedrichs - lewy ( cfl ) condition becomes @xmath3 , where @xmath4 is the parameter measuring scale separation .
standard explicit methods become useless for spdes with multiple scales .
such time - step restriction can in theory be removed by using implicit methods as was shown in @xcite .
however the implicitness of the numerical scheme forces one to solve potentially large linear algebraic problems at each time step .
furthermore , it was shown in @xcite that implicit methods are not suited for studying the long time dynamics of fast - slow stochastic systems as they do not capture the correct invariant measure of the system .
although this result has been obtained for finite dimensional stochastic systems , it is expected that it also applies to infinite dimensional fast - slow systems of stochastic differential equations ( sdes ) , rendering the applicability of implicit methods to spdes with multiple scales questionable .
we also note that a new class of explicit methods , the s - rock methods , with much better stability properties than the euler - maruyama method was recently introduced in @xcite .
although these methods are much more efficient than traditional explicit methods , computing time issues will occur when trying to solve spdes with multiple scales as considered here , since the stiffness is extremely severe for small @xmath5 furthermore , capturing the correct invariant measure of the spde for @xmath6 is still an issue for such solvers . in this paper
we consider spdes of the form @xmath7 posed in a bounded domain of @xmath8 with appropriate boundary conditions .
the differential operator @xmath9 is assumed to be a non - positive self - adjoint operator in a hilbert space @xmath10 , @xmath11 denotes space - time gaussian white noise , @xmath12 is the covariance operator of the noise and we take @xmath4 .
we assume that the operator @xmath9 has a finite dimensional kernel , @xmath13 .
this assumption leads to scale separation between the slow dynamics in @xmath13 and the fast dynamics in the orthogonal complement of the null space @xmath14 where @xmath15 . in this paper
we will furthermore assume that noise acts directly only on the orthogonal complement @xmath16 . when noise acts also on @xmath13 , different distinguished limits than the ones considered in this paper should be considered . in order to describe the longtime behavior of the spdes we perform an _ advective _ rescaling set @xmath17 .
using the scaling properties of white noise we obtain the following singularly perturbed spde @xmath18 another scaling is of interest , namely the _ diffusive _ rescaling @xmath19 which leads to the spde @xmath20 for concreteness , we will focus on the class of spdes with quadratic nonlinearities that was considered in @xcite , and assume that @xmath21 where @xmath22 is a quadratic function ( e.g. @xmath23 , a symmetric bilinear form ) , @xmath24 a linear function and the exponent @xmath25 is either @xmath26 or @xmath27 . and @xmath24 involve derivatives of the function @xmath28 .
for example , for both the burgers and the kuramoto - shivashinsky equation we have @xmath29 . the linear function @xmath30 is included to induce a linear instability to the dynamics . in the case of the burgers equation
we will simply take @xmath31 whereas in the case of the kuramoto - shivashinsky equation we can take @xmath32 . further discussion can be found in section [ sec : numerics ] and in @xcite . ]
the choice of @xmath25 will depend on the particular scaling .
singularly perturbed spdes with quadratic nonlinearities provide a natural testbed for testing the applicability of the heterogeneous multiscale method to infinite dimensional stochastic systems , since a rigorous homogenization theory exists for this class of spdes @xcite .
furthermore , spdes of this form arise naturally in stochastic models for climate @xcite and in surface growth @xcite .
finally , it has already been shown through rigorous analysis and numerical experiments that these systems exhibit a very rich dynamical behavior , such as noise - induced transitions @xcite and the possibility of stabilization of linearly unstable modes due to the interaction between the additive noise and the scale separation @xcite .
we believe , however , that the methodology developed in this paper has a wider range of applicability and is not restricted to spdes with quadratic nonlinearities .
further comments about the class of spdes for which we believe that the proposed numerical method can be applied can be found in section [ sec : conclusions ] .
our numerical algorithm is based on a combination of a spectral method with micro - macro time integration schemes .
we denote by @xmath33 the projection onto @xmath13 and by @xmath34 , @xmath35 the projection onto @xmath16 .
we then rewrite and as fast - slow system of sdes [ e : fast_slow_intro_adv ] @xmath36 and [ e : fast_slow_intro_diff ] @xmath37 where the functions @xmath38 and @xmath39 are the projections of @xmath40 onto @xmath13 and @xmath16 .
we remark that an @xmath41 nonlinear term can be added in . the fast - slow systems and resemble fast slow systems for sdes (
however , the fast process @xmath42 is infinite dimensional and the well known averaging and homogenization theorems @xcite do not apply .
averaging and homogenization results for spdes have been obtained recently @xcite . in particular , provided that the fast process @xmath42 in has suitable ergodic properties , then the slow variable @xmath43 converges , in the limit as @xmath44 tends to @xmath45 to the solution of the averaged equation @xmath46 where the averaged coefficient is given by the average of @xmath47 with respect to the invariant measure of the ( infinite dimensional ) fast process @xmath42 .
when this average vanishes ( i.e. the centering condition from homogenization theory is satisfied ) then the dynamics at the advective time scale becomes trivial and it is necessary to look at the dynamics at the diffusive time scale , equations .
it was shown in @xcite that the slow variable @xmath43 of this system of equations , the solution of , converges in the limit as @xmath44 tends to @xmath48 to the solution of the homogenized equation @xmath49 with explicit formulas for the homogenized coefficients see section [ sec : multisc ] for details . for finite dimensional fast systems ,
the coefficients in and can be calculated , in principle , in terms of appropriate long - time averages see @xcite for details .
the numerical method proposed in @xcite and analyzed in @xcite , coined under the name of the heterogeneous multiscale method ( hmm ) , relies on the numerical approximation of the coefficients in and by solving the original fine scale problem on time intervals of an intermediate time scale and use that data to evolve the slow variables using either or . in this paper
we show how this methodology , when combined with a spectral method , can also be applied to spdes with multiple scales , that is , to the systems and .
the aim of the present paper is to present the algorithm and report numerical experiments .
the analysis of the proposed numerical method and the extension to more general classes of spdes with multiscale structure will be presented in a forthcoming paper .
the rest of the paper is organized as follows . in section [ sec : algorithm ] we present our new algorithm .
analytical and computational techniques for the analysis of spdes with multiple scales at the heart of the multiscale algorithm are presented in [ sec : multisc ] . in section [ sec : numerics ] we present numerical experiments .
section [ sec : conclusions ] is reserved for conclusions and discussion on future work .
we propose a numerical algorithm to approximate numerically the solution of based on a micro - macro algorithm , capable of capturing the effective behavior of the spde .
we explain the numerical algorithm for the case of diffusive time scale ( the hardest numerically ) and comment on the advective time scale later in this section .
we consider spdes in a hilbert space @xmath10 with norm @xmath50 and inner product @xmath51 .
@xmath9 denotes a differential operator , @xmath11 space - time white noise and @xmath12 the covariance operator of the noise .
we assume that @xmath9 is a self - adjoint nonpositive operator on @xmath10 with compact resolvent .
we denote its eigenvalues and ( normalized ) eigenfunctions by @xmath52 : @xmath53 the eigenfunctions of @xmath9 form an orthonormal basis in @xmath10 .
we assume that @xmath9 and the covariance operator of the noise @xmath12 commute .
thus , we can write , formally , @xmath54 where @xmath55 are independent one - dimensional white noise processes , i.e. , mean - zero gaussian processes with @xmath56 . here
@xmath57 and @xmath58 are the usual kronecker delta functions .
furthermore , we will assume that @xmath9 has a finite dimensional kernel @xmath59 , @xmath60 and write @xmath61 we introduce the projection operators @xmath62 and write @xmath63 , @xmath64 .
finally , we will assume that noise acts only on @xmath16 , i.e. @xmath65 . * step 1 .
decomposition in a fast - slow system . * + using the projection operators defined in and , equation can be written as a fast - slow stochastic system [ e : x - y ] @xmath66 where @xmath67 since @xmath68 we order the pairs of eigenfuctions and eigenvalues such that the kernel @xmath13 is spanned by the first @xmath69 eigenfunctions of @xmath9 .
we can write @xmath70 moreover , we introduce @xmath71 as mentioned in the introduction we will often consider the case @xmath72 .
then the above decomposition reads @xmath73 where we notice that for a linear function @xmath74 we simply have @xmath75 .
then , in view of and we can rewrite the system in the form [ e : x - y-1 ] @xmath76 equations , resp . , are the infinite system of singularly perturbed sdes that we want to solve numerically .
+ * step 2 .
* + we consider a finite dimensional truncation of the above system and keep @xmath77 fast processesinstead of @xmath78 and similarly for the eigenvalues @xmath79 and the noise intensity @xmath80 . ] : [ e : x - y-1_fin ] @xmath81 where @xmath82 and @xmath83 and @xmath84 and @xmath85 . for the decomposition ,
, we will use the notations @xmath86 where @xmath87 and @xmath88 with components similar as in or .
+ * step 3 .
numerical solution of the reduced system . *
+ the reduced system is solved by a micro - macro algorithm following @xcite .
it consists of a macrosolver ( we use the notation @xmath89 ) , chosen here to be the euler - maruyama scheme @xmath90 where @xmath91 ( the wiener increment ) is @xmath92 .
notice that @xmath1 represents here a macrotime step , i.e. , @xmath1 can be chosen much larger than @xmath44 .
the drift function @xmath93 and diffusion function @xmath94 appearing in , recovered from a time - ensemble average , are given by @xmath95 where @xmath96 are the solutions of a suitable auxiliary system ( given in below ) involving the fast problem . here
@xmath97 denotes the number of samples taken for the numerical calculation , @xmath98 the number of micro timesteps and @xmath99 a number of initial micro timesteps that are omitted in the averaging processes to reduce the effect of transients ( see below ) . *
auxiliary system . *
as observed in @xcite , for diffusive timescales , computing effective coefficients via time - averaging ( relying on ergodicity ) , may require to solve over time @xmath100 . to overcome this problem , it was suggested again in @xcite to replace the fast process in by ( @xmath101 ) [ e : auxiliary ] @xmath102
the numerical approximations @xmath103 of and , respectively , are the functions appearing in the averaging procedure to recover the macroscopic drift and diffusion functions ( see ) - ) .
notice that we fix the slow variables in the system at the current macro state @xmath104 .
we use again the euler - maruyama method and compute @xmath103 as [ e : microsolver ] @xmath105 where @xmath106 and @xmath107 is a @xmath108 random variable .
the index @xmath109 refers to the macrotime , @xmath110 .
we compute over @xmath111 microtime steps , over @xmath112 microtime steps to compute the time - ensemble average . notice that for the microsolver , the timestep @xmath113 resolves the fine scale @xmath114 .
the initial values for the micro solver are taken to be ( for @xmath115 @xmath116 and @xmath117 for @xmath118 .
the motivation for computing the above time averages is given in the next section .
we notice that the auxiliary system is degenerate , since the noise in is additive .
this implies that the results presented in ( * ? ? ?
b ) are not applicable in this case and a more elaborate analysis is required for proving geometric ergodicity .
this analysis , based on the ergodic theory for hypoelliptic diffusions @xcite , will be presented elsewhere . in the present paper
we will assume that the auxiliary process is ergodic .
[ [ advective - time - scale .
] ] advective time scale .
+ + + + + + + + + + + + + + + + + + + + + a similar algorithm can be derived for the advective time scale .
we consider the fast - slow system that after projection and truncation reads [ e : x - y-1_fin_ad ] @xmath119 similarly as .
the macrosolver , chosen to be the euler explicit method , is given by @xmath120 where the effective force @xmath121 is given by the time average @xmath122 where @xmath123 is a numerical approximation of the truncated fast system with a slow variable fixed at time @xmath110 .
as previously , @xmath97 denotes the number of samples and @xmath112 the number of micro timesteps and @xmath99 is the number of initial micro timestep ommited to reduce the transient effects . for the advective
scaling , there is no need for an auxiliary problem for the micro solver @xcite .
in this section we summarize recent results on the averaging and homogenization for spdes @xcite that are the analytical foundation on which the numerical algorithm presented in section [ sec : algorithm ] is built . in this section
we briefly discuss the analytical form of the effective system corresponding to . under the assumption that the vector field @xmath124 ( see ) is centered with respect to the invariant measure of the fast process
, @xmath125 then the slow process converges to a homogenized equation of the form @xmath126 where @xmath127 represent an @xmath128dimensional wiener process and the sde is interpreted in the it sense .
the subscript @xmath77 are used to emphasise the fact that the homogenized coefficients depend on the number of fast processes that we take into account . an analytic expression for the coefficients that appear in
is given by [ e : coeff_homog ] @xmath129 here @xmath130 denotes the invariant measure of the process @xmath131 which is given by and @xmath132 denotes the invariant measure of the the process @xmath133 .
notice that @xmath134 is the solution of the rescaled process corresponding to , i.e. , @xmath135 .
alternatively , the calculation of the coefficients @xmath136 and @xmath137 which appear in the homogenized equation can be obtained by the solution of the poisson equation @xmath138 where @xmath139 is the generator of the fast ( truncated ) ornstein - uhlenbeck process .
this process is ergodic and its invariant measure is gaussian : @xmath140 where @xmath141 denotes the normalization constant .
we notice that the system is a finite dimensional fast - slow system of sdes for which standard homogenization theory applies @xcite . for quadratic nonlinearities
the poisson equation can be solved analytically .
the calculation of the coefficients in the homogenized ( amplitude ) equation reduces then to the calculation of gaussian integrals that can also be done analytically .
this will be done in section [ sec : diff_time_scale ] .
averaging problems for fast - slow systems of spdes were studied recently in @xcite and their results can be applied to .
one important observation is that in the system , the fast process is , to leading order @xmath142 , an infinite dimensional ornstein - uhlenbeck process .
the ergodic properties of such an infinite dimensional process can be analyzed in a quite straightforward way and the invariant measure , if it exists , is a gaussian measure in an appropriate hilbert space that can be written down explicitly @xcite .
assuming that the process @xmath143 is ergodic with gaussian invariant measure @xmath144 with mean @xmath48 and covariance operator @xmath145 , then the slow process @xmath43 converges to the solution of the averaged equation @xmath146 where the integration is over an appropriate hilbert space .
when @xmath147 in is given in terms of a symmetric bilinear map , i.e. , @xmath148 the calculation of the vector field that appears in the averaged equation reduces to the calculation of gaussian integrals and can be performed explicitly . in this case
we have @xmath149 where @xmath150 and where we used the notation @xmath151 and @xmath152 denotes the dimension of the null space of @xmath9 .
then , the fast - slow system becomes [ e : averag_quadratic ] @xmath153 and the averaged equation for reads @xmath154 where @xmath155 in the case when the null space is one - dimensional , @xmath156 , the averaged equation becomes @xmath157 with @xmath158 and @xmath159 .
this equation can be solved in closed form : @xmath160 we remark that solutions to , depending on the choice of the initial conditions , do not necessarily exist for all times .
we also remark that it is straightforward to consider the case where there is an additional higher order linear term ( in @xmath44 ) in the equation , i.e. @xmath161 . in this case
the averaged equation becomes @xmath162 where @xmath163 .
we consider the system obtained after a diffusive time rescaling to . in order to describe the homogenized equation
, we further assume that @xmath164 in is of the form @xmath165 where @xmath166 is a symmetric bilinear map satisfying @xmath167 .
we recall that the noise does not act directly on the slow variables , @xmath168 , where @xmath69 is the dimension of the null space of @xmath9 . under appropriate assumptions on the quadratic nonlinearity and on the covariance operator of the noise , together with the assumptions on @xmath9 and @xmath12 stated earlier in this section , it is possible to prove @xcite that the projection of the solution to onto the null space of @xmath9 , @xmath169 , converges weakly to the solution of the homogenized sde ( the amplitude equation ) @xmath170 where the noise is interpreted in the it sense and the drift @xmath171 given by @xmath172 where the linear map @xmath173 and the trilinear map @xmath174 are defined by @xmath175 in the above we used the notation @xmath176 and @xmath177 , whereas @xmath178 stands for the symmetric tensor product we denote by @xmath179 its symmetric tensor product .
similarly , we use the notation @xmath180 for the symmetric tensor product of two elements and @xmath181 for the symmetric tensor product of two linear operators .
furthermore , we extend the bilinear form @xmath182 to the tensor product space by @xmath183 . more details can be found in ( * ? ? ?
4 ) . ] and where we have defined @xmath184 the quadratic form associated with the diffusion matrix @xmath185 is given by @xmath186 furthermore , the fast process can be approximated by an infinite dimensional ornstein - uhlenbeck process . the precise statement and proof of the above results
can be found in @xcite .
the assumption that the @xmath187 term in is linear is needed in order to go from to after rescaling or , equivalently , to .
if our starting point is the already rescaled spde , then we can apply the results from @xcite to nonlinearities of the form @xmath188 where @xmath189 is an arbitrary nonlinearity . in this case
the drift term in the amplitude equation becomes @xmath190 where @xmath191 denotes the invariant measure of the fast ou process .
when the null space of @xmath9 is one dimensional and , consequently , the homogenized sde is a scalar equation , it is possible to obtain sharp error estimates and to prove convergence in the strong topology . in this case equation
becomes @xmath192 where @xmath193 in the one dimensional case the formulas for the coefficients that appear in the homogenized equation have a simpler form than in the multidimensional case .
in particular , we have , with @xmath194 : [ e : coeffs ] @xmath195 it is worth mentioning that if we are using a non - orthonormal basis , i.e. a basis @xmath196 , then the coefficients that appear on the right hand side of the above equation transform according to @xmath197 we also have @xmath198 .
the formulas for the coefficients that appear in the amplitude equation can be also obtained by writing the spde in fourier space , truncating and then using singular perturbation theory - type of techniques for the corresponding backward kolmogorov equation @xcite .
more details on this approach can be found in @xcite .
we also remark that , in general , both additive as well as multiplicative noise will appear in the amplitude equation , although only ( degenerate ) additive noise is present on the spde .
in this section we apply our numerical method to several spdes and report its convergence and performance .
we consider here several examples of spdes with quadratic nonlinearities and check that the theory developed in @xcite and summarized in section [ sec : multisc ] applies .
for all of these examples we can derive rigorously the homogenized equation , with explicit formulas for the coefficients and therefore , we can present a rigorous numerical study for our algorithm and test the effectiveness of the proposed numerical algorithm .
we will consider variants of the burgers and the kuramoto - shivashinsky ( ks ) equations ( with a linear instability term added ) in one dimension with either dirichlet or periodic boundary conditions . in particular , we will consider the singularly perturbed spdes ( i.e. we have already rescaled to the diffusive time scale ) @xmath199 and @xmath200 respectively , where the noise @xmath11 is as in section [ sec : multisc ] .
the operator @xmath12 , the covariance operator of the noise , has eigenvalues @xmath201 and eigenfunctions @xmath202 , which are also the eigenfunctions of the differential operator that appears in either or , i.e. the two operators commute .
we will consider these two equations either on @xmath203 $ ] with dirichlet boundary conditions or on @xmath204 $ ] with periodic boundary conditions .
for the burgers nonlinearity and for the boundary conditions that we consider it is straightforward to check that the centering condition @xmath167 is satisfied . a more natural equation to consider than would be the ks equation in the small viscosity regime , i.e. @xmath205 where @xmath206 .
this equation can be rewritten in the form @xmath207 the theory presented in @xcite and the numerical scheme developed in this paper apply to this equation .
the application of the numerical method developed in this paper to equation and to related models will be presented elsewhere .
some recent analytical and numerical results on the behaviour of solutions to have been reported in @xcite .
we will use the notation @xmath208 it is possible to check that for the above equations and for the chose boundary conditions the theory developed in @xcite and summarized in section [ sec : multisc ] applies . consider first equations and on
@xmath203 $ ] with dirichlet boundary conditions . in this case
the null space of @xmath209 and @xmath210 is one dimensional : @xmath211 with @xmath212 being either @xmath182 or @xmath213 .
the normalized eigenfunctions of @xmath209 and @xmath210 are @xmath214 .
the corresponding eigenvalues are @xmath215 since the null space is one - dimensional , the homogenized equation is a scalar sde . for the nonlinearity @xmath216 = \frac{1}{2 } \partial_x ( u v ) $ ] it is straightforward to calculate @xmath194 .
we have @xmath217 where @xmath218 denotes the kronecker delta .
we can then use formulas to calculate the formulas that appear in the homogenized equation .
let @xmath219 of either @xmath220 or @xmath210 with dirichlet boundary conditions .
the homogenized equation is given by that we recall here for convenience @xmath221 where @xmath222 the coefficients that appear in can be computed as and use formula . ]
[ e : coeffs_homog_burgers ] @xmath223 in the case where only the second mode is forced with noise , @xmath224 then the coefficients become @xmath225 in this case only multiplicative noise appears in the homogenized equation and it can lead to intermittent behavior of solutions as well as noise induced transitions @xcite .
we will also consider either the burgers or the ks equation on @xmath204 $ ] with periodic boundary conditions . in this case
the null space of both @xmath209 and @xmath210 is two - dimensional and is spanned by @xmath226 with @xmath212 being either @xmath182 or @xmath213 .
the homogenized equation is given by , where @xmath227 .
it consists of a system of two coupled sdes .
we can use formulas and , together with the formula for the nonlinearity @xmath216 = \frac{1}{2 } \partial_x ( u \ , v)$ ] to calculate the coefficients that appear in the homogenized equation .
we shall now apply our numerical algorithm to the model problems , described in section [ sec : examples ] . as
the behavior of our algorithm is similar for the burgers and the kuramoto - shivashinsky equation we will do a thorough numerical study on the burgers equation and comment on the results for the kuramoto - shivashinsky equation .
* burgers equation .
* we consider equation on @xmath203 $ ] with homogeneous dirichlet boundary conditions .
we know from section [ sec : examples ] that , for @xmath44 sufficiently small , we have that @xmath228 where @xmath229 is the solution of .
the function @xmath230 in depends on @xmath231 which for the burgers equation can be computed using formulas with @xmath232 .
they read @xmath233 following the algorithm described in section , we look for a solution to of the form @xmath234 substitute the expansion in to obtain the a fast - slow system of sdes as described in .
following the algorithm of section [ sec : algorithm ] we compute numerically the slow variable @xmath104 as @xmath235 where @xmath236 are given by and , respectively .
we also consider the truncated homogenized problem , i.e. , @xmath237 where @xmath238 and where @xmath239 are obtained from , with the sums truncated at @xmath77 . for numerical comparison
we also compute @xmath240 the euler - maruyama approximation of the sdes and , respectively . the same brownian
path will be used in , and .
we emphasize that the numerical solutions for and rely on analytically computed homogenized coefficients , whereas for we implement the multiscale algorithm of section [ sec : algorithm ] , where the coefficients @xmath241 are computed `` on the fly '' and rely on the microsolver and .
hence no a - priori analytical knowledge of the amplitude equation is required .
we choose the values of the various parameters entering in the averaging process for the computation of @xmath242 as suggested in @xcite , i.e. , @xmath243 , @xmath244 . according to @xcite , this guarantees ( for the case of non - degenerate fast processes ) that the error is bounded by @xmath245 .
in our case with a degenerate fast process an error bound is still to be established .
here we monitor such convergence numerically .
more precisely , we set @xmath246 and monitor the error using @xmath247 for various values of @xmath248 , where @xmath249 and @xmath250 represent the final time .
notice that captures the error between the homogenized solution of the truncated system and the numerical solution of the truncated system , while , where the index @xmath251 stands for limit , captures the error between the homogenized solution of the limit problem and the numerical solution of the truncated system .
* 2-mode truncation .
* we set @xmath252 in and substitute the expansion in to obtain the following system of equations [ e:3mode ] @xmath253 the auxiliary process can be derived as explained in section [ sec : algorithm ] and reads [ e : auxiliary3 ] @xmath254
we apply the algorithm of section [ sec : algorithm ] to get a numerical approximation of the homogenised problem corresponding to .
the final time is @xmath255 and @xmath256 , which corresponds to macro time - step of size @xmath257 .
the macro solver for the method is given by .
as mentioned above , we compare our results with and .
the unknown coefficients @xmath258 in can be computed using and , where the sum is truncated at @xmath259 and read @xmath260 .
.,width=226 ] we observe in figure [ fig3mode ] that we get numerically the expected order of convergence corresponding to @xmath261 .
furthermore , as the micro time - step becomes smaller , the numerical scheme gets closer to and slightly deviates from .
this is expected as the numerical solution is not converging to that latter solution .
we observe nevertheless that with only two fast modes , the numerical scheme already captures quite well the effective behavior of the slow variable of the infinite dimensional system .
we also illustrate the time evolution of one trajectory comparing over the time @xmath262 with @xmath263 , the euler - maruyama method for the amplitude equation , the homogenized equation and the macro solver . the same brownian path is used for generating the three trajectories and as well as the same macro time step . ) , ( solution denoted @xmath264 ) and ( solution denoted @xmath265 ) for three paths ( left @xmath266 for @xmath104 , middle @xmath267 for @xmath104 , right @xmath268 for @xmath104 ) .
we use 2-mode truncation for and .,title="fig:",width=170,height=151 ] -8 mm ) , ( solution denoted @xmath264 ) and ( solution denoted @xmath265 ) for three paths ( left @xmath266 for @xmath104 , middle @xmath267 for @xmath104 , right @xmath268 for @xmath104 ) .
we use 2-mode truncation for and .,title="fig:",width=170,height=151 ] -8 mm ) , ( solution denoted @xmath264 ) and ( solution denoted @xmath265 ) for three paths ( left @xmath266 for @xmath104 , middle @xmath267 for @xmath104 , right @xmath268 for @xmath104 ) .
we use 2-mode truncation for and .,title="fig:",width=170,height=151 ] we perform this comparison for increasing accuracy of the micro solver used to recover the macro data , namely , @xmath269 we see in figure [ fig : trajectory ] that the trajectory for the amplitude equation and the homogenized equation coincide , while the macro solver gets closer to the true dynamics as we refine the micro time step . for the same trajectory we also give a space - time plot for the approximation of the original spde @xmath270 with @xmath229 solution of the amplitude equation , the homogenized equation or the macro solver .
again we see that the numerical method captures the right behavior of the solution . of the spde ; @xmath271 ( left figure @xmath266 ) , @xmath272 ( middle figure ) and @xmath273 ( right figure).,width=529,height=264 ] * 3-mode truncation . * we set @xmath274 in and obtain the following system of equations [ e:4mode ] @xmath275 the auxiliary process can be computed similarly as for the 3-mode truncation .
we perform the same set of numerical experiments as for the 3-mode truncation , reported in figure [ fig4mode ] .
similar behavior than previously noted can be observed .
observe that the discrepancy between the numerical scheme and gets smaller .
this is expected as with additional modes , the homogenized equation ( that we aim at approximating with our multiscale scheme ) gets closer to . .
, width=226 ] [ fig4mode ] * 4-mode truncation . * we set @xmath276 in and apply the similar procedure as previously . for the sake of brevity , we do not write the system of equations in this case and just report the numerical convergence . we see in figure [ fig5mode ] a similar behavior of our numerical scheme as observed previously .
we again notice that the discrepancy between the numerical scheme and is smaller than for lower order truncation .
notice that the first numerical result reported is obtained for @xmath277 with @xmath267 .
this is due to stability issues with the euler - maruyama scheme for the fast process . as the linear term in the equation for the fifth mode
@xmath278 is @xmath279 , the stability restriction @xmath280 implies @xmath281 and thus the time step @xmath282 is unstable . .
, width=226 ] [ fig5mode ] + [ [ the - kuramoto - shivashinsky - equation . ] ] the kuramoto - shivashinsky equation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the equations for the @xmath77-mode truncation of the kuramoto - shivashinsky equation are very similar to the ones for the burgers equation and will not be presented here .
the only difference is that the fast process is more dissipative than for the burgers equation , due to the stronger dissipativity of the operator @xmath210 compared to @xmath220 .
as the results of the numerical experiments for the ks equation are very similar to the results reported in this section for the burgers equation , they will not be presented here .
we have presented a new numerical method for the efficient and accurate solution of stochastic partial differential equations with multiple scales .
the new numerical scheme is based on a combination of a spectral method with the hmm methodology and has been tested on spdes with quadratic nonlinearities for which a rigorous homogenization theory exists .
this enables us to check the performance of our method .
the numerical experiments presented in this paper suggest that the new method performs well and allows to solve accurately multiscale spdes by solving a low dimensional fast - slow system of sdes .
the method is suitable for infinite dimensional stochastic systems for which there is clear separation of scales , and for which a low dimensional homogenised ( or averaged ) equation for the slow modes exists .
there are still many questions that are left open .
first , the rigorous analysis of the proposed method and a careful study of the convergence and stability properties of the proposed method remains to be done .
in addition , the optimisation of the proposed method by tuning appropriately the parameters of the algorithm has not been performed yet .
this appears to be an open problem even when the hmm methodology is applied to finite dimensional fast / slow systems of sdes @xcite .
the proposed numerical algorithm could be used to study in detail the qualitative and quantitative properties of solutions to spdes with quadratic nonlinearities , since spdes of this form exhibit very rich dynamical behaviour .
furthermore , we would like to apply the numerical algorithm to more general classes ( and systems ) of semilinear spdes , for which an averaged or homogenised equation is known to exist .
examples include systems of reaction / diffusion equations that were considered in @xcite as well as the swift - hohenberg spde @xcite . in our algorithm
, we did not make use of the fact that the form of the amplitude equation ( i.e. a landau equation with additive and multiplicative noise ) is known .
knowledge of the functional form of the coefficients that appear in the homogenised or averaged equation can be used in order to simplify the numerical algorithm .
the stochastic landau equation appears as the amplitude equation for several infinite dimensional stochastic dynamical systems , not only for spdes with quadratic nonlinearities , e.g. @xcite .
thus , the algorithm proposed in this paper , could be modified to develop an efficient method for studying the dynamics of spdes near bifurcation .
all these topics are currently under investigation .
part of this work was done while gp was visiting the mathematics section of epfl .
the hospitality of the department and of the group of a. abdulle is greatly acknowledged .
the authors thank s. krumscheid for an extremely careful reading of an earlier version of the paper and for many useful remarks .
gp is supported by the epsrc .
_ analysis of multiscale methods for stochastic dynamical systems with multiple time scales , _ siam mul .
, 8:944 - 964 , 2010 .
a. j. majda , ch .
franzke , and b. khouider .
an applied mathematics perspective on stochastic modelling for climate .
, 366(1875):24292455 , 2008 .
g. c. papanicolaou , d.w .
stroock , and s. r. s. varadhan
. martingale approach to some limit theorems . in _
papers from the duke turbulence conference ( duke univ . ,
durham , n.c . , 1976 ) ,
paper no . 6 _ , pages ii+120 pp .
duke univ .
duke univ . ,
durham , n.c . ,
1977 .
m. pradas , d. tseluiko , s. kalliadasis , d. t. papageorgiou , and g. a. pavliotis .
noise induced state transitions , intermittency , and universality in the noisy kuramoto - sivashinksy equation .
, 106(6):060602 , feb 2011 . | a new method for solving numerically stochastic partial differential equations ( spdes ) with multiple scales is presented .
the method combines a spectral method with the heterogeneous multiscale method ( hmm ) presented in [ w. e , d. liu , and e. vanden - eijnden , comm .
pure appl .
math .
, 58(11):15441585 , 2005 ] . the class of problems that we consider are spdes with quadratic nonlinearities that were studied in [ d. blmker , m. hairer , and g. a. pavliotis , nonlinearity , 20(7):17211744 , 2007 . ] for such spdes an amplitude equation which describes the effective dynamics at long time scales can be rigorously derived for both advective and diffusive time scales .
our method , based on micro and macro solvers , allows to capture numerically the amplitude equation accurately at a cost independent of the small scales in the problem .
numerical experiments illustrate the behavior of the proposed method .
stochastic partial differential equations ; multiscale methods ; averaging ; homogenization ; heterogeneous multiscale method ( hmm ) |
network coding attracted the attention in wireless networks @xcite .
the network coding benefit is easily represented by reducing the number of transmissions . for example , in two - way relay networks @xcite of figure [ f : twoway ] , the traditional method requires four slots to exchange data , whereas the network coded relaying requires three slots @xcite . in @xcite ,
moreover , two - slot transmission strategy is studied . in the example
, the relay node should have data from both source nodes to execute network coding .
this means the source nodes must transmit their data before the relay node does .
consequently , scheduling can significantly impact the performance of network coding .
the packet scheduling in multiuser communications was originally motivated by @xcite , where only one user that has the best channel condition should transmit to achieve the capacity . since then , many scheduling algorithms have been proposed @xcite . in @xcite ,
the authors proposed a cross layer scheme that integrates per - hop packet scheduling , network coding and congestion control . to maximize coding opportunities , they investigated the scheduling of transmissions .
however , they considered only time invariant case , where all links have the identical channel gain . in @xcite ,
the authors considered a throughput optimal scheduling problem with random traffic .
they showed that both digital and analog network coding outperform the plain routing . in @xcite
, the authors pointed that if network coding and scheduling are designed separately , the expected throughput gain may not be achieved .
when it comes to throughput maximization criterion , the buffers of nodes are assumed full .
the full buffer assumption is not realistic though it may provide tractableness of analysis .
the network situation can arbitrarily change over time by node mobility , channel fading and the amount of data . considering these dynamics
, throughput may not be well defined . instead
, the time it takes to transfer a bunch of data volume at each source can be practical measurement of the network performance @xcite . in this point of view
, we consider the finite amount of initial data traffic by introducing a _ time span minimization _ criterion @xcite .
the details of time span minimization are explained later .
scheduling in time varying channels is a difficult problem . in @xcite , the opportunistic scheduling , which exploits time varying nature of the radio environment ,
is proposed .
scheduling algorithms with consideration of energy - efficiency @xcite , limited channel information @xcite are investigated . in this paper
, we propose a scheduling scheme incorporating network coding and channel varying information for the two - way relay networks .
the provided channel information will constrain the choice of the scheduling policy . if we know the current and future instantaneous channel state information , we can adjust appropriate transmission rates . however , in practical situations , we know only the current channel information and the distribution of the channel fluctuation .
our question is how to design scheduling algorithms incorporating network coding and channel varying information .
we consider three channel models , time invariant , time varying with finite states and rayleigh fading .
the first two models may less reflect practical situations than the rayleigh fading model .
however , we can obtain an optimal solution by analysis in the first two cases , which helps us understanding the relation between scheduling and network coding .
in rayleigh fading model , we propose a heuristic algorithm and obtain meaningful results by simulations . throughout the paper ,
we use the time span as the performance metric for our scheduling algorithm .
the time span means the amount of time needed to send all the data in every source node in the network .
the problem was first defined in uplink packet scheduling for the cellular cdma system @xcite . by minimizing the time span
, we can shorten the end - to - end delay while keeping the per - node throughput reasonably high .
the rationale behind is to take care of both throughput and delay .
the rest of the paper is organized as follows . in section 2 ,
we describe the system model .
our scheduling scheme for the minimum time span is proposed in section 3 . in section 4 ,
we verify the performance of the proposed scheme by simulations .
section 5 concludes the paper .
consider the two - way relay networks as shown in figure [ f : twoway ] .
the source nodes at both ends communicate with each other using the relay node in the middle .
the relay node carries out network coding ( e.g. , exclusive or ) operations to the received data from two source nodes .
@xmath0 is the maximum data rate through the link @xmath1 during a time slot .
@xmath0 is constant , but randomly varies between slots due to the time varying channel .
the channel model between the two nodes is reciprocal .
then , the link is symmetric in that both directions have the same rate .
we consider two models for the time varying channel model .
the first one is a finite channel state model .
each data rate varies in a set of finite possible rates @xmath2 . without loss of generality , we assume @xmath3 .
we represent the channel state as a vector of data rates @xmath4 . in a time slot ,
the channel state is one of possible combinations of data rates .
for example , if @xmath5 , then there are four channel states : @xmath6 , @xmath7 , @xmath8 and @xmath9 .
we denote the probability that the channel state is @xmath10 as @xmath11 . in a time slot , if the _ assumed _ channel state vector (
i.e. , the rates using a specific level of modulation and coding ) is element - wisely less than or equal to the _ actual _ channel state vector , then transmissions succeed .
the second model is the continuous state model of rayleigh fading .
each channel power gain follows an exponential distribution with a mean value .
let @xmath12 denote the finite amount of data that source node @xmath1 wants to send to the other source node .
we assume only one node can transmit at any time instant , i.e. , tdma .
to propose a scheduling scheme for the minimum time span , we start with the time invariant channels , and extend the result to the case of randomly varying channels .
it is a special case of the time varying channels ; one of @xmath11 s is one , and the others are zero . to begin with ,
consider the rate region that depicts the amount of _ transmittable _ data in a single time slot ( see figure [ f : varying_rate_region1 ] ) .
the rate region @xcite is a convex set and the network coding is adopted by the relay node . in the figure , the corner point @xmath13 represents the data rate of the ( network coded ) three - mode transmission . in the first fraction of a time slot , a source node sends data to the relay node . in the second fraction ,
the other source transmits . in the last fraction ,
the relay node broadcasts the network coded data to both ends .
note that @xmath14 is the maximum achievable rate by network coding to exchange data between source nodes .
the network coded data rates of both end nodes are equal .
this is because the network coded data rates are symmetric and bounded by the minimum of maximum transmission rates of links .
note that @xmath15 is the maximum achievable rate by one - directional forwarding from source node @xmath16 to the other source node .
two edge points @xmath17 and @xmath18 represent one - directional forwarding .
the line segment between the corner point ( @xmath19 ) and an edge point ( @xmath20 or @xmath21 ) denotes the time multiplexing of network coding and one - directional forwarding . from theorems 3.1 and 3.2 of @xcite ,
the maximum rates of one - directional forwarding and network coding are : @xmath22 to minimize the time span , we formulate the minimum time span into the following linear programming problem : @xmath23 where @xmath24 denotes the transmission time of source node @xmath25 to source node @xmath26 by one - directional forwarding , and @xmath27 corresponds to that of source node @xmath26 . the time for network coded transmission
is denoted by @xmath28 .
the maximum network coded data size is @xmath29 because the relay node codes the same amount of received data from both source nodes .
the objective is to minimize the total transmission time to send the target amount of data @xmath30
. * proposition 1 . * _ to achieve the minimum time span , the time for network coding operation should be @xmath31 .
_ * proof .
* we can rewrite the first constraint of in terms of @xmath32 as follows : @xmath33 then objective function of can be rewritten as : @xmath34 without loss of generality , we assume @xmath35 and from : @xmath36 because @xmath37 , it is @xmath38 . to minimize the objective function , @xmath32 should be as large as possible .
therefore , @xmath39 the consequence of proposition 1 is summarized as follows : first , either node 1 or node 2 transmits its data during @xmath40 or @xmath41 , respectively .
then , the relay node transmits the received data with network coding during @xmath42 and with the one - directional forwarding for flushing the remaining data during @xmath43 or @xmath44 .
proposition 1 explains what the solution of the time span minimization looks like but a question still remains regarding how to achieve it .
we propose a rate region based scheduling to achieve the minimum time span .
we draw a line segment across @xmath45 and @xmath30 as in figure [ f : varying_rate_region ] .
the drawn line has the slope @xmath46 .
there is a cross point @xmath47 between the rate region and the line . the number @xmath48 denotes the data rate of node @xmath16 under the channel state @xmath1 . to minimize the time span
, each source node should transmit @xmath49 and @xmath50 under the channel state @xmath1 , by time multiplexing of network coding and one - directional forwarding .
then a time slot is over .
we draw another line between @xmath45 and the remaining data @xmath51 and repeat the same procedure .
this intuitively explains how the minimum time span scheduling works .
consider the channels vary randomly with channel states . for example , if the @xmath5 , we can draw four instantaneous rate regions as in figure [ f : varying_rate_region ] .
nodes hardly find out the current channel state , but know the probability @xmath52 that the channel state is @xmath10 . due to the randomness of the channel , the _ expected _
time span will depend on the scheduling policy .
if a source node selects the rate region of @xmath53 , the transmission succeeds only when the actual rate region is @xmath53 .
however , if the source node selects the rate region of @xmath54 , its transmission succeeds for all channel realizations .
we want to minimize the expected time span . for the purpose ,
let us consider a probabilistic scheduling policy such that the probability to select @xmath10 is @xmath55 .
the expected data rate of source node @xmath16 in @xmath56 is defined as @xmath57 , where @xmath58 is the index set of transmittable channel states @xmath1 .
@xmath59 $ ] denotes the expectation of an instantaneous data rate of node @xmath16 . for two vectors @xmath60 and @xmath61
, we define operator @xmath62 if @xmath60 is element - wisely less than or equal to @xmath61 .
similar to the time invariant case , the ratio of expected data rates @xmath63/e[r_1]$ ] should be @xmath64 to minimize the time span .
we can derive expected data rate of node @xmath25 as : @xmath65 = e_{q_i } \left [ { e\left [ { r_1 = 1}^{n^2 } { \left ( { \sum\limits_{j \in a^{\left ( i \right ) } } ^{n^2 } { p_j r_1^{\left ( i \right ) } } } \right)q_i } .\ ] ] we assume finite amount of data that source nodes want to send . in this case , the time span minimization problem can be solved by solving expected data rate maximization problem . to this end , we consider the following optimization problem : @xmath66 where @xmath67 is the sum of expected data rates of source nodes 1 and 2 when channel state @xmath1 is selected . in , the linear programming is solved by assigning 1 to @xmath55 that has the largest coefficient : @xmath68 each node selects the channel state using @xmath69 and decides its transmission rate .
even if the channel fluctuates randomly , each source transmits / schedules its data as if the channel were fixed .
consequently , the time for network coded transmission can be find by proposition 1 .
the above scheduling configuration is performed at the beginning of each time slot to decide transmission order and data rate .
the node that decides scheduling policy ( the relay node is favorable ) must know initial data size of the other nodes .
due to the nature of two - way relay networks , the data size information of the other source nodes is readily available .
we can apply the following protocol to share data size information .
first , relay node requests source nodes to inform their data sizes .
source nodes send data size information to the relay node , respectively . after gathering data size information
, the relay node computes the transmission rates of sources by .
it broadcasts these scheduling information to sources .
we assume now that channels experience the rayleigh fading .
the channel power gain @xmath70 varies between slots following the exponential distribution with a known parameter @xmath71 . in this case
, the approach in the previous subsections is inapplicable and analytical results may not be found .
we introduce a total power constraint @xmath72 ( w ) , the additive white gaussian noise power @xmath73 ( w / hz ) and focus on the relay node operation .
we assume that the relay node has received all the data from sources . according to proposition 1 , to minimize time span
, the relay node should have at least @xmath74 data from each of sources .
then the problem is reduced to the power allocation problem over slots .
the water filling is known to be optimal for allocating powers to orthogonal multi - channels in throughput maximization .
we have the channel information _ causally _ if the past and current channel information are available . on the other hand
, we have the channel information _ noncausally _ if the future channel information is known . in the noncausal channel information case
, we can regard each time slot as a orthogonal subchannel block .
if the number of time slots is fixed , we can solve the throughput maximization problem by the water filling power allocation .
given the fixed amount of data , the time span is minimized by throughput maximization at the relay node .
if the amount of received data from both sources is equivalent ( @xmath75 ) , the relay node can always transmit network coded data .
we consider the following optimization problem for relay node : @xmath76 where , @xmath77 is the number of required time slots to transmit the given data and @xmath78 is channel power gain for link @xmath1 at time slot @xmath79 .
the total throughput over the time slots should be greater than or equal to the received data size to flush out the data .
we propose to solve the problem in the following way .
the procedure is also depicted in figure [ f : wf_noncausal ] .
first , assuming @xmath80 , allocate the power budget @xmath72 to the first time slot .
if the total throughput over the slot is greater than or equal to the remaining data size , a feasible solution is founded .
if not , increase one time slot @xmath81 , and , apply the water filling allocation again with the power budget @xmath72 .
we continue this procedure until flushing out the remaining data .
figure [ f : wf_noncausal_-5dbm ] is an example of the power allocation .
the total power constraint is -5 dbm and the data size is 30 mbytes .
five slots are consumed to transmit all data . in practice ,
the above algorithm may not be applicable because the future channel knowledge can not be obtained . on the other hand ,
if we know the channel distribution and its mean value , the following heuristic approach can be used instead .
the algorithm is described in figure [ f : wf_causal ] .
first , assuming @xmath80 and a virtual time slot @xmath82 , then virtually apply the water filling allocation with power budget @xmath72 .
if the total throughput over the time slots is greater than or equal to the remaining data size , a feasible solution is founded .
if not , increase one virtual time slot @xmath83 , assume that the future channel is the same as the mean value of the channel distribution and virtually apply the water filling allocation again .
when flushing out the remaining data , the power @xmath84 allocated the current time slot , is actually used .
recalculate the power budget and the remaining data size .
we continue this procedure until the relay node has no remaining data .
figure [ f : wf_causal_-6dbm ] is an example of the heuristic power allocation .
total power constraint is -6 dbm and data size is 15 mbytes .
total 26 time slots are consumed to transmit the given data .
the water level is not flat because actual channel information is not given .
however , the power is allocated to relatively high channel quality slots .
this means the heuristic power allocation works like the water filling .
the heuristic algorithm can be combined with one - directional forwarding . before the power allocation step ,
the relay node compares the throughput of three cases , i.e. , network coding and two one - directional forwarding transmissions , and selects the transmission scheme shown the largest throughput at each time slot .
the other type of incorporation is also possible .
for example , we can put high priority on the network coding .
if network coding is possible at the current time slot , it is selected even though one - directional forwarding shows higher throughput . in this way , a heuristic algorithm combined with one - directional forwarding can be applied for the case that the data sizes from two sources are different .
the performance of this algorithm is verified in following section .
we simulated our heuristic power allocation algorithms under the rayleigh fading channel .
we assume that channel power gains follow the exponential distribution with mean 1 .
the amounts of data to transmit are @xmath85 mbytes .
channel bandwidth is 100 mhz .
the noise power is fixed to @xmath86 w / hz . figure [ f : wf_tsm ] shows the average time span as a function of the power budget . for comparison , noncausal and
causal one - directional forwarding cases are plotted together .
to execute one - directional forwarding , the relay should decide to whom it should transmit , i.e. , the relay node should select one of the two links .
the relay node chooses the highest quality channel between the two channels . with noncausal channel knowledge ,
the relay node can select future channels in advance .
after that , the relay node allocates power using water filling . in the causal case ,
the relay node can select only one of current channels and allocates power with mean valued future channel gain . as the power budget increases ,
the average time span decreases for all the cases .
network coding outperforms one - directional forwarding in all cases considered . the performance gap becomes bigger under a small power budget . with -10 dbm power budget ,
the gap between the network coding with causal channel information and one - direction forwarding with causal channel information is about 2.2 slots .
however , with 0 dbm power budget , the gap is about 1 .
this result shows that network coding improves the power efficiency .
the network coding with the causal channel information , i.e. , only executing network coding at the relay node , shows almost the same performance with the network coding with the noncausal channel information as the power budget increases .
however , the opportunistic network coding with causal channel information , i.e. , selecting proper one between network coding and one - direction forwarding at the current time slot , shows worse performance than the network coding with causal channel information .
this result means that network coding opportunity should be maximized to minimize the time span .
the imbalance between two buffers can be occurred by the opportunistic network coding and it reduces network coding opportunity . to further investigate the relation between time span minimization and network coding , we simulated opportunistic network coding , network coding first scheme ( high priority on the network coding ) and one - directional forwarding with various data ratio ( @xmath87 ) .
@xmath88 is fixed to 17 mbytes . due to the imbalance between buffers ,
one - directional forwarding should be used to flush the remaining data .
figure [ f : wf_tsm_data_ratio ] shows the average time span as the data ratio varies . when the data ratio is zero , network coding can not be applied and one - directional forwarding is the only option .
when the data ratio becomes positive , one - directional forwarding ( solid line with circle marker ) shows poor performance because the relay node should transmit the both direction . in all cases ,
the network coding first scheme ( solid line with triangle marker ) outperforms the opportunistic network coding ( solid line with square marker ) .
the results mean that maximizing the network coding opportunity is a better strategy than maximizing the throughput of the current slot .
we proposed a scheduling scheme incorporating network coding and channel varying information for the network coded two - way relay networks . in time varying channels with finite states , to minimize the expected time span , the proposed scheduler operates as if channels stay in the specific channel state .
even if the channel experiences a constant mean valued rayleigh random fading , the average time span can be minimized when the relay node transmits / schedules network coded data as much as possible .
a possible research direction is to find an optimal network coding scheduling under the rayleigh fading channel and to extend the results , considering other network coding strategies ( e.g. , physical - layer network coding ) .
this research was supported by basic science research program through the national research foundation of korea ( nrf ) funded by the ministry of education , science and technology ( 2009 - 0088483 ) . | we investigate a scheduling scheme incorporating network coding and channel varying information for the two - way relay networks .
our scheduler aims at minimizing the time span needed to send all the data of each source of the network .
we consider three channel models , time invariant channels , time varying channels with finite states and time varying rayleigh fading channels .
we formulate the problem into a manageable optimization problem , and get a closed form scheduler under time invariant channels and time varying channel with finite channel states . for rayleigh fading channels , we focus on the relay node operation and propose heuristic power allocation algorithm resemblant to water filling algorithm . by simulations , we find that even if the channel fluctuates randomly , the average time span is minimized when the relay node transmits / schedules network coded data as much as possible . network coding , scheduling , time span , two - way relay networks . |
high energy processes can be described within quantum chromodynamics ( qcd ) through its factorization properties .
these allow us to write the cross sections by separating the hard part , which is the well defined interaction of a quark with _
e.g. _ a virtual photon , from the soft part , described in terms of parton distributions functions ( pdfs ) .
the latter can not be calculated from first principles .
information about the pdfs can only be obtained directly from experiment .
several collaborative groups in the past decades have sought to use fitting techniques to study the pdfs behavior by devising models / analytic forms for the various pdfs , @xmath0 , @xmath1 , at light cone momentum fraction @xmath2 , and scale @xmath3 , whose parameters are constrained by experiment .
more recently , new types of partonic distributions that improve considerably our ability to study the nucleon structure and its partons dynamics became accessible experimentally .
the new distributions extend , in different ways , the concept of pdfs to observables that can measure the spin , spatial and momentum correlations in hadrons .
these are the transverse momentum distributions , tmds ref . , and the generalized parton distributions , gpds ( see review in ref . ) .
for gpds and tmds one increases the range of kinematical variables that need to be measured by introducing both transverse momenta , and the fourier conjugates of the partonic spatial degrees of freedom . while for the pdfs one faces the problem of considering different results / extractions that vary significantly from group to group , lacking a well defined theory to constrain them ( for a review of recent results see _ e.g. _ refs . ) , for tmds and more crucially for gpds , standard global analyses are doubtful due to the increased number of variables compared to the scarcity of the experimental data ( see _ e.g. _ ref . for a state of the art analysis ) .
a relatively new approach to pdfs fitting is the one proposed by the nnpdf collaboration ref .
, who have replaced the standard analytic forms of pdfs with a more complex neural network ( nn ) solution .
the estimated uncertainties for nnpdf fits are larger than those of global fits , possibly indicating that the global fit uncertainties might be underestimated . in ref .
a criticism was put forward about relying on purely automated fitting procedures such as the ones used by nnpdf . a new specific type of neural network , the self - organizing map ( som ) , was proposed .
the main point is that since for nnpdfs the effect of modifying individual nn parameters is unknown , the result might not be under control in the extrapolation region , or in between the data points if the data are sparse .
this issue is even more important when extending the fitting procedures to a wider set of semi - inclusive and exclusive observables .
we therefore pushed forward with the som method , and we improved the preliminary work in ref . by restructuring the original code in such a way that on one side , a fully quantitative error analysis can be implemented , and on the other we now have sufficient flexibility to allow for analyses of different observables , including the matrix elements for deeply virtual exclusive and semi - inclusive processes .
our first quantitative results for the unpolarized case using next - to - leading - order ( nlo ) perturbative qcd in the @xmath4 scheme were presented in ref .. in this workshop we also discuss a possible procedure to extend soms to the extraction of gpds .
recent years have seen an outstanding growth in the usage of neural networks as paradigms for computational methods .
soms are a type of neural network developed by t. kohonen in the 80s ref . based on a topological mapping of the external environment onto the brain s internal neural connections . in soms
the nodes / neurons map cells are tuned to a set of input signals / data / samples according to a form of adaptation .
the various nodes form a topologically ordered map during the learning process .
this feature constitutes perhaps the main strength of soms in that it allows one to obtain a simplified two - dimensional representation of complex data otherwise depending on a " hard to control set of parameters ( see figure [ fig0 ] ) . in this respect
soms can be considered as a non linear form of principal component analysis which is often used to analyze highly dimensional data .
another aspect that sets soms apart from other nns is their learning process .
soms learn via _ unsupervised learning _ whereas the learning of generic artificial nns is _
supervised_. in supervised learning a set of examples is given , and the goal is to force the data to match the examples as closely as possible . a cost function is defined that measures the importance to miss or detect the correct result . during the learning process
the cost function is minimized . in unsupervised learning
the cost function is minimized without introducing a definite set of examples , but just by similarity relations , or by finding how the data cluster or self - organize .
theoretical studies of the mechanism for map evolutions are in progress . in phenomenology , many new uses , one of which described below , might derive from this property .
soms are built as two dimensional arrays whose cells get sensitized / tuned to a specific set of input signals according to a given order . to illustrate this in fig.[fig0 ]
we show the simple example of a `` color map '' , where the input signals are represented by the three colors to be associated with each one of the map cells .
the som algorithm consists of three stages : _
i ) _ initialization ; _ ii ) _ training ; _ iii ) _ mapping . square map , and its input signals represented by colors .
the arrows indicate the signal to cell matching procedure . in our more realistic situation ,
the map and signals will constitute of either pdfs or dis structure functions.,scaledwidth=35.0% ] during the initialization procedure weight vectors of dimension @xmath5 are associated to each cell @xmath6 : @xmath7\ ] ] @xmath8 are given spatial coordinates , _
i.e. _ one defines the geometry / topology of a 2d map that gets populated randomly with @xmath8 . for the training , a set of input data @xmath9 , \ ] ] ( isomorphic to @xmath8 )
is then presented to @xmath8 , or compared via a `` similarity metric '' that we choose to be @xmath10 the most similar weight vector is the best matching unit ( bmu ) .
soms are based on unsupervised and `` competitive '' learning .
this means that the cells that are closest to the bmu activate each other in order to `` learn '' from @xmath11 . practically , they adjust their values according to @xmath12 , \label{learn_1}\ ] ] where @xmath5 is the iteration number , and @xmath13 is the `` neighborhood function '' defining a radius on the map which decreases with both @xmath5 , and the distance between the bmu and node @xmath6 . in our case
we use square maps of size @xmath14 , and @xmath15 where @xmath16 is the number of iterations . at the end of a properly trained som , cells that are topologically close to each other will contain data which are similar to each other . in the final phase
the actual data are distributed on the map and clusters emerge .
( note that the specific location of the clusters on the map is not relevant and will not necessarily be the same from one run to another ; only the clustering properties are important . )
since each map vector now represent a class of similar objects , the som is an ideal tool to visualize high - dimensional data , by projecting it onto a low - dimensional map clustered according to some desired similarity feature .
in ref.ref . a som algorithm was constructed which , together with a genetic algorithm ( ga ) was applied to pdf fitting . in this initial work
it was proven that the som method works well as a minimization technique , in that it was shown that the @xmath17d.o.f . decreased towards unity with each ga iteration .
however problems connected with the smoothing of the functions remained , related to the stochastic procedure used to generate them and to initialize and train the map . these prevented a fully quantitative analysis of the data , including uncertainty evaluations . a solution to this problem
was obtained in the present work ref . by developing a new version of the sompdf code where several important improvements were introduced . by writing it in a single compiled language
fortran 95 the code was made more flexible and faster since a parallel ( mpi ) version was easily implemented . as explained in more detail below , the increased flexibility and speed allow us to perform now random variations on the parameters of the various input pdfs , instead of on the pdf values themselves , thus providing continuous solutions which are amenable to standard error analyses .
our ultimate goal is to provide a procedure to extract the gpds , and their related observables , including _
e.g. _ the spatial partonic d.o.f . from experimental data .
we therefore added flexibility in the code s inputs so as to accommodate the additional kinematical variables and parameters .
we now describe the main fitting procedure , originally applied to pdfs .
this first phase of our work involves only marginally the clustering and visualization properties of soms . for pdfs
this is , in fact , barely needed since the inclusive scattering databases provide a rather large kinematical coverage , and only two kinematical variables are necessary to describe the data .
soms are used mostly to realize random variations of the pdf curves .
our fitting procedure is based on the initialization , training , and mapping steps described for a general case in section [ sec2 ] . in a nutshell , a set of database / input pdfs is formed by selecting at random from a range of existing pdf functions and varying their parameters .
baryon number and momentum sum rules are imposed at every step .
these input pdfs , evolved to nlo to the desired @xmath3 , are used to initialize the map .
a subset of input pdfs is then used to train the map .
notice that this phase of the procedure implies defining specific criteria for selecting the pdfs that will define each subset .
two possible criteria are discussed in ref .. the similarity is tested by comparing the pdfs , according to eqs.([learn_1 ] ) and ( [ learn_2 ] ) at given @xmath2 and @xmath3 values .
the new map pdfs are obtained by averaging the neighboring pdfs with the bmu pdfs . once the map is trained a ga is implemented .
the @xmath18 per input pdfs is calculated with respect to the experimental data .
we then take a subset of these functions with the best @xmath18 , and use them as seeds to form a new set of input pdfs .
we train the map with the new set of input pdfs and repeat the process .
the @xmath18 was found to decrease monotonically towards @xmath19 with every ga iteration ( fig.[fig1 ] ) .
our stopping criterion is established when the @xmath18 stops varying
its curve flattens .
/d.o.f for one of our sompdf runs.,scaledwidth=45.0% ] map for a set of 43 runs .
the panels represent ( clockwise from upper left)the @xmath20 , @xmath21 , @xmath22 , @xmath23 distributions , respectively , at @xmath24 gev@xmath25 .
the shaded areas are our results including the error analysis outlined in the text . for comparison
we show also results from several other parametrizations .
lower panel : @xmath20 , @xmath23 and gluon distributions at @xmath26 gev@xmath25 , along with their variations obtained for @xmath17 d.o.f .
@xmath27 , from our previous sompdf analysis . left : @xmath28 map , right : @xmath29 map .
these figures are shown to illustrate the improvement in the curves smoothing attained in our new procedure ( adapted from ref.).,scaledwidth=75.0% ] map for a set of 43 runs .
the panels represent ( clockwise from upper left)the @xmath20 , @xmath21 , @xmath22 , @xmath23 distributions , respectively , at @xmath24 gev@xmath25 .
the shaded areas are our results including the error analysis outlined in the text . for comparison
we show also results from several other parametrizations .
lower panel : @xmath20 , @xmath23 and gluon distributions at @xmath26 gev@xmath25 , along with their variations obtained for @xmath17 d.o.f .
@xmath27 , from our previous sompdf analysis . left : @xmath28 map , right : @xmath29 map .
these figures are shown to illustrate the improvement in the curves smoothing attained in our new procedure ( adapted from ref.).,scaledwidth=75.0% ] our first runs , presented in figure [ fig2 ] , used a `` test '' set of data from dis consistent with the sets used in refs .. the data sets chosen were from bcdms , h1 , nmc , slac and zeus ( see _ e.g. _ references in ref . ) .
the shaded areas in the figure represent our error bands . in our preliminary
runs we defined a statistical error on an ensemble of sompdf runs .
finally , with a fully working procedure in hand , we comment on our capability to extend the pdf analysis to the gpd case .
here one has a total of @xmath30 observables per parton component , given by the real and imaginary part of the compton form factors ( cffs ) containing the corresponding gpds ref .. the number of observables that has been defined _
e.g. _ in ref .
is @xmath31 of which @xmath32 appear at leading order .
an open question is therefore to establish which experiments , observables , and with what precision they can determine the various gpd components .
our analysis will provide a step in this direction in that by applying the clustering properties of the soms we will be able to quantitatively discern the various cffs , that will become `` dimensions '' in our analysis , and to establish the sensitivity of the different experiments to each one of them .
an illustration of how clustering of gpds on the som might be represented is given in fig.[fig3 ] .
more work is in progress ref ..
in this work we described a new computational method based on self - organizing maps for parametrizing nucleon pdfs .
future developments will also be directed at exploiting the full potential of soms that offer the capability of going beyond a fully automated procedure , by enabling one to control the fitting procedure at each step .
the selection of the best pdf candidates for the subsequent iteration could then be made based on the user s preferences instead of solely based on the @xmath18 .
our program can be extended to multivariable cases such as the generalized parton distributions where the data are too sparse for stochastically generated , parameter - free , pdfs .
we thank the university of virginia alliance for computational science and engineering for computer time , and the hpc group at jefferson lab , in particular david richards and chip watson , for allotting us space on their clusters .
this work is partially supported by the u.s .
department of energy grant de - fg02 - 01er4120 ( s.l and d.z.p . ) . | we present a new method to extract parton distribution functions from high energy experimental data based on a specific type of neural networks , the self - organizing maps .
we illustrate the features of our new procedure that are particularly useful for an anaysis directed at extracting generalized parton distributions from data .
we show quantitative results of our initial analysis of the parton distribution functions from inclusive deep inelastic scattering . |
the concept of casimir forces,@xcite i.e. long - ranged effective interactions between macroscopic bodies immersed in strongly fluctuating media , is nowadays recognized in plethora of systems ranging from biology to cosmology .
on one hand their prominent importance stems from providing possibilities of testing theories of fundamental interactions,@xcite on the other they are invoked to explain important mechanisms underlying physical phenomena in organic matter@xcite and as a control tool for nanodevices.@xcite the more conventional and well - studied ( also experimentally ) examples of casimir forces in condensed matter include fluid mixtures@xcite and helium.@xcite the essence of the phenomenology underlying the casimir forces is the presence of soft fluctuations occurring within a medium , whose spectrum of excitations becomes constrained by the boundary conditions imposed by the confining walls or by macroscopic objects immersed therein . as a result ,
the free energy acquires a contribution depending on the separation @xmath7 between the walls ( or the objects ) and the system finds it favorable to either increase or decrease this distance .
a prominent characteristic of casimir forces is the universality of its asymptotic behavior ( including the amplitude ) for @xmath8 and @xmath9 , where @xmath10 is a system - specific microscopic lengthscale and @xmath11 denotes the ( bulk ) correlation length for the medium fluctuations .
if the system features spontaneously broken continuous symmetry , one should additionally distinguish the transverse and longitudinal bulk correlation lengths .
the system then displays a crossover between the goldstone and critical regimes upon varying temperature .
for a recent study of this crossover see ref . .
on the other hand , the amplitude of the force ( even its sign ) does depend on the boundary conditions .
the orthodox qed casimir force@xcite is a pure effect of quantum vacuum fluctuations .
its theoretical description involves a free field theory , where interactions with matter are taken into account via boundary conditions and the expectation value of the fluctuating electromagnetic field is zero .
the situation is quite distinct in these respects in many condensed matter systems . here
a field - theoretic description of the problem often requires a treatment of interaction terms in the effective action . for certain cases
, the boundary terms may also lead to nonzero expectation values of the ( nonuniform ) order parameter , which gives rise to effective interactions between the confining walls ( or other macroscopic objects ) also in the absence of fluctuations . in this work
we analyze the critical casimir effect employing a class of standard models to describe different aspects of critical behavior , namely the @xmath1-dimensional @xmath0-symmetric @xmath12-type models .
the system is confined to be finite in one direction ( so that its extension is @xmath7 ) and we consider the periodic boundary conditions . despite not being of the highest relevance for quantitative comparison with experimental data , this choice of boundary conditions does not yield casimir forces at mean - field level , so that the effective interaction is a pure fluctuation effect ( preserving the analogy to the qed setup ) .
in addition , these boundary conditions yield a zero - mode in the free propagator , which leads to substantial complications in a conventional rg treatment , where the @xmath13-expansion@xcite becomes ill - defined beyond two - loop order.@xcite we study the system within the one - particle - irreducible ( wetterich ) framework of functional rg.@xcite the exact flow equation governing the flow of the generating functional for irreducible vertices is treated within an approximation retaining the flow of interaction couplings of arbitrary order , but neglecting the renormalization of the momentum dependencies in the propagators .
this amounts to the leading order of an approximation strategy known as derivative expansion,@xcite which was successfully applied _
inter allia _ for accurate calculations of the @xmath0 bulk critical behavior@xcite ( including the kosterlitz - thouless case@xcite ) .
this approach is not controlled by any small parameter , but may be considered as a leading order of a systematic procedure to compute critical properties .
it can be applied for different @xmath1 and @xmath14 . in particular , we scan the dependence of the critical casimir force amplitude @xmath2 as a function of dimensionality @xmath1 , varying @xmath1 between two and three at fixed @xmath15 .
our findings compare very well to the exact value@xcite for @xmath4 and @xmath5 and reasonably well with numerical monte carlo ( mc ) simulations@xcite in @xmath6 ( both for @xmath5 and @xmath16 ) .
the paper is organized as follows : in sec . ii we present the model and the rg setup . in sec .
iii we consider the bulk critical behavior upon varying @xmath1 between two and three at fixed number of order - parameter components @xmath17 .
in particular , within the applied approximation , we compute the correlation length in the vicinity of the critical point and extract the values of the critical exponent @xmath18 , describing the divergence of the correlation length @xmath11 at criticality . in the subsequent sec .
iv we present the computation of the critical casimir forces in the cases @xmath5 and @xmath16 , continuously varying @xmath1 . in sec .
ivc we gauge the accuracy of the applied exact rg truncation by performing a refined calculation including the wave function renormalization in the case @xmath5 , @xmath4 .
we show that such a refinement has a negligible impact on the obtained casimir forces .
we summarize the work in sec .
the landau - type model considered here is a paradigm of the theory of phase transitions and critical phenomena . in particular , in refs .
it was employed to study the thermal casimir forces .
the fluctuating @xmath14-component real order - parameter field @xmath19 is confined in a @xmath1-dimensional box of volume @xmath20 , where @xmath21 is much larger than all the other lengthscales present in the system .
this imposes an idealization as compared to the systems realized in experiments and simulations , where the aspect ratio @xmath22 is always finite . for a discussion of the aspect ratio dependencies of casimir forces
see refs . .
we impose periodic boundary conditions in all the directions and the effective hamiltonian takes the standard form @xmath23=\int_v d^dx[u_0(\rho)+\frac{1}{2}z_0(\nabla\vec{\chi})^2]\ ; , \label{lgw}\ ] ] where we introduced @xmath24 , and the effective potential @xmath25 is quadratic in @xmath26 .
the model involves the interaction coupling @xmath27 , the mass - like parameter @xmath28 , the constant @xmath29 , and is additionally supplemented with a lower cutoff in real space at @xmath10 , which we treat on equal footing with the other parameters .
the parameter @xmath28 may be tuned by varying temperature , pressure or some other physical quantity . at mean - field level
its critical value is @xmath30 . on the other hand ,
accounting for fluctuation effects favors the disordered phase and reduces this value , so that @xmath31 .
the present approach@xcite relies on an exact rg equation governing the flow of the scale - dependent effective action @xmath32 $ ] , where the field @xmath33 is the expectation value of @xmath19.@xcite upon decreasing the momentum cutoff scale @xmath34 from @xmath35 to zero , the functional @xmath36 $ ] interpolates between the initial action @xmath37 $ ] at @xmath38 and the full effective action , i.e. the free energy @xmath39 $ ] at @xmath40 .
the quantity @xmath32 $ ] may be understood@xcite as the free energy obtained after including the fluctuation modes with momentum @xmath41 $ ] . at @xmath38 no fluctuations are included and @xmath32 $ ] is just the bare effective action .
on the other hand , in the limit @xmath42 all the fluctuation modes are included .
the flow of @xmath32 $ ] upon reducing the cutoff scale @xmath34 is governed by the flow equation@xcite
@xmath43=\frac{1}{2}\tilde{\mathrm{tr}}\frac{\partial_\lambda r_\lambda ( q^2)}{\gamma^{(2)}_\lambda[\vec{\phi}]+r_\lambda ( q^2 ) } \ ; , \label{wetterich_eq}\ ] ] where @xmath44 is a momentum cutoff function added to the inverse propagator to regularize the infrared behavior , and @xmath45=\delta^2\gamma_{\lambda}[\vec{\phi}]/(\delta \phi^a_{\vec{q}}\delta \phi^a_{-\vec{q}})$ ] , with @xmath46 ( no off - diagonal components of the inverse propagator occur in the present problem ) .
the trace @xmath47 sums over field components and momenta : @xmath48 in the limit @xmath49 and with @xmath50 , @xmath51 .
integration of eq .
( [ wetterich_eq ] ) with the initial condition at @xmath35 specified by eqs .
( [ lgw ] ) and ( [ initial_u ] ) yields the exact solution to the model .
herein we apply an approximation strategy known as derivative expansion,@xcite where the symmetry - allowed terms in @xmath32 $ ] are classified according to increasing number of derivatives and we keep only the leading ( zeroth ) order term in this expansion .
this truncation level ( the so - called local potential approximation ( lpa ) ) amounts to imposing the following ansatz : @xmath52 = \int_v d^dx[u_\lambda(\rho)+\frac{1}{2}z_0(\nabla\vec{\phi})^2]\ ; , \label{lpa_ansatz}\ ] ] where we put @xmath53 .
the present approximation therefore neglects renormalization of all the momentum dependencies in the propagator , but does not truncate interaction terms in the local effective potential @xmath54 at any order . in sec .
ivc we perform an additional calculation capturing the wave function renormalization for @xmath4 and @xmath5 .
our results indicate that the effect of such a truncation refinement on the casimir forces is negligible even though it significantly improves the quality of the obtained bulk critical exponents . with the ansatz ( [ lpa_ansatz ] ) the effective action flow
is represented by the flow of the scale - dependent effective potential @xmath55\right\}\ ; , \label{eff_pot_flow}\ ] ] where @xmath56 and @xmath57 .
since we neglect the renormalization of @xmath29 , eq .
( [ eff_pot_flow ] ) is closed and our subsequent analysis is based on its numerical integration .
any practical calculation requires specifying the cutoff function @xmath58 , which must respect a number of conditions.@xcite we make the following choice,@xcite known as the litim cutoff @xmath59 for which the trace in eq .
( [ eff_pot_flow ] ) can be straightforwardly performed in the limit @xmath60 .
the free energy is identified as @xmath61 = k_b t v u_{\lambda\to 0}(\rho)\;,\ ] ] where the functional is evaluated at a uniform field configuration .
its dependence on @xmath7 arises due to the trace involving the @xmath7-dependent momenta allowed by the boundary conditions .
the casimir forces , in turn , result from the @xmath7-dependent free energy . since our study concerns the long - ranged interactions close to a second - order phase transition , we first need to resolve the critical behavior , which is done in the next section .
in order to efficiently tune the system to its bulk criticality , we consider the limit @xmath60 and perform the standard variable transformation @xmath62 where the superscript @xmath63 indicates that we deal with an infinite system .
this brings eq .
( [ eff_pot_flow ] ) to a scale - invariant form @xmath64\ ; , \;\;\ ; \label{u_flow}\end{aligned}\ ] ] where @xmath65 the primes denote differentiation with respect to the argument .
we introduced @xmath66 and @xmath67 , where @xmath68 is the surface area of a @xmath1-dimensional unit sphere .
we numerically integrate the discretized version of eq .
( [ u_flow ] ) taking eq .
( [ initial_u ] ) as the initial condition .
this is done by considering a grid in the @xmath69-space .
the discrete @xmath69-derivatives were computed numerically using a 5-point routine at each rg step . for accuracy control
we varied the grid size between @xmath70 and @xmath71 points .
this discretization procedure therefore maps the partial differential eq .
( [ u_flow ] ) onto a set of @xmath72 coupled nonlinear ordinary differential equations .
this system can be handled relying on the standard available numerical stepping procedures including the simplest euler - type routines .
for all of the analysis of the present section we put @xmath73 and @xmath74 . at fixed @xmath1 and the ( initial ) interaction coupling @xmath27 in eq .
( [ initial_u ] ) , we scan the solution to eq .
( [ u_flow ] ) for @xmath75 upon varying the bare mass parameter @xmath28 in eq .
( [ initial_u ] ) .
we identify two regimes in the space spanned by @xmath76 , corresponding to the symmetric and the symmetry - broken phases .
the line of critical points ( in the @xmath76 space ) and the corresponding fixed points are found by the dichotomy procedure illustrated in fig . 1 . .
the present exemplary plot corresponds to @xmath77 , @xmath5 , and @xmath78 . for @xmath79
the plotted flowing coupling @xmath80 attains positive values at a finite scale , indicating flow to the symmetric phase . for @xmath81
the coupling @xmath80 decreases upon reducing the scale @xmath34 as the system flows to the symmetry - broken phase .
fixed - point behavior represented by the plateau corresponds to the critical value @xmath82 .
, width=321 ] having identified the critical point , we vary @xmath28 so that the system remains in the symmetric phase close to @xmath83 and scan the behavior of the ( non - rescaled ) effective potential at its minimum in the limit @xmath75 . by computing the curvature of the effective potential as function of @xmath84 at @xmath85 for different values of @xmath79 in the vicinity of the critical point , we extract the correlation length @xmath86 .
the obtained behavior very well fits the expected power - law @xmath87 we analyze the dependence of the exponent @xmath88 focusing on @xmath15 and varying @xmath1 between two and three .
the obtained dependencies are shown in figs 2 and 3 for @xmath5 and @xmath16 , respectively .
plotted as function of system dimensionality @xmath1 at @xmath5 .
the value @xmath89 agrees well with the best available estimates . for @xmath1 approaching 2 the obtained result substantially deviates from the exact value @xmath90 .
see the main text for a discussion of this discrepancy.,width=321 ] plotted as function of system dimensionality @xmath1 at @xmath16 .
the value @xmath91 compares well to the best available estimates .
we observe a rapid growth of @xmath92 upon reducing @xmath1 toward 2 .
this tendency is in line with the expectation that the @xmath93 model in @xmath4 features an essential singularity of the correlation length when approaching criticality from the symmetric phase.,width=321 ] in both cases the values of @xmath94 compare well to the best available estimates@xcite in @xmath6 .
we obtain @xmath95 and @xmath96 ( the best estimates according to ref . are @xmath97 and @xmath98 up to two digits ) . on the other hand ,
our results for @xmath5 and @xmath1 approaching 2 substantially deviate from the exact value @xmath90 .
this discrepancy arises because the present exact rg truncation neglects renormalization of the momentum dependencies in the propagator .
indeed , the singularity of the propagator renormalization at criticality is quantified by the value of the anomalous dimension @xmath99 , which is relatively small in @xmath6 ( @xmath100 for @xmath5 ) , but sizable ( @xmath101 for @xmath5 ) for @xmath4 .
the present approximation , eq .
( [ lpa_ansatz ] ) , obviously yields @xmath102 .
this also perturbs the very structure of eq .
( [ u_flow ] ) at @xmath4 , where the vanishing factor @xmath103 should be replaced by @xmath104 . for this reason we did not achieve true fixed point behavior at @xmath4 and do not present the value of @xmath94 precisely for @xmath4 ( see fig.2 ) .
this however does not create any obstacle in computing the casimir forces in @xmath4 and @xmath5 close to the critical point in the symmetric phase ( see the next section ) .
we readdress the case @xmath4 , @xmath5 with a refined approximation capturing @xmath105 in sec .
ivc . for the same reason we are not able to resolve the critical behavior of the kosterlitz - thouless universality class ( @xmath4 , @xmath16 ) , where nonzero @xmath99 is crucial to capture the physics even at a qualitative level .
interestingly , for @xmath16 , we observe a rapid increase of @xmath92 upon reducing @xmath1 towards 2 .
this may indicate the divergence of @xmath92 and the onset of the kosterlitz - thouless physics , where the correlation length exhibits an essential singularity in the symmetric phase .
we also note , that the computed dependence @xmath106 nicely fits a power law @xmath107 in the interval @xmath108 $ ] .
we obtain @xmath109 .
we also note , that the values of @xmath18 in low @xmath1 significantly improve upon including @xmath99 , i.e. going to higher order in derivative expansion . for a calculation resolving the @xmath1-dependence for @xmath5 at the next - to leading order in derivative expansion see ref . .
one may at this point ask , to which extent the inaccuracy in resolving the bulk critical behavior in @xmath110 will affect the results for the casimir forces , the amplitudes in particular . at the outset one possibility
is that the computed properties simply come out inaccurate .
many critical quantities are however insensitive to the value of @xmath99 . in the present context , for example , it is known that the critical casimir forces decay asymptotically as @xmath111 irrespective of the value of @xmath99 .
the asymptotic behavior of the correlation function ( involving @xmath99 ) therefore does not directly influence the power law governing the decay of the casimir force .
another example one may invoke here is the universal exponent @xmath112 governing the shape of the finite - temperature transition line in quantum critical systems.@xcite despite being located in the portion of the phase diagram which is dominated by non - gaussian thermal fluctuations , @xmath112 is fully determined by the system dimensionality @xmath1 and the dynamical exponent @xmath113 .
the precise values of @xmath99 or @xmath94 are therefore not relevant to it .
the results of the subsequent section suggest that the casimir amplitudes also belong to this category .
indeed , our calculation yields a very accurate estimate of the critical casimir amplitude in @xmath4 , where bulk critical exponents are off the correct values and overall the approximation is less reliable than in the case @xmath114 . to support the above statement , and the conclusions of our computations , in sec .
ivc we perform an additional calculation including the flow of @xmath29 in @xmath4 , @xmath5 .
this cures the drawbacks of the above analysis for @xmath60 , leads to significant improvement of the bulk critical exponents , but , as turns out , has negligible impact on the obtained casimir forces .
having resolved the critical behavior for the infinite system , we now consider finite @xmath7 and proceed by computing the casimir forces .
the system state is chosen in the symmetric phase ( @xmath79 ) , but close enough to the critical point , so that the correlation length @xmath115 . the @xmath7-dependent contribution to the free energy resulting in the appearance of the casimir force is identified as @xmath116 where @xmath117 this leads to @xmath118\;. \label{sigma}\ ] ] the additional superscript in the effective potential @xmath119 indicates the finite size dependence that we set out to compute now .
the potential @xmath120 is evaluated at @xmath121 , because , as indicated above , we perform the computations in the symmetric phase .
utilizing eq .
( [ eff_pot_flow ] ) with the cutoff given in eq .
( [ litim_cutoff ] ) , we have @xmath122 , \label{u_casi1}\ ] ] and @xmath123\times \nonumber\\ \times\left[1 + 2\sum_{n=1}^{n_{max}}\left(1-\left(\frac{2\pi n}{\lambda l}\right)^2\right)^{\frac{d-1}{2}}\right ] \;,\;\;\;\;\ ; \label{u_casi2}\end{aligned}\ ] ] where @xmath124 and the quantities @xmath125 are given by eq .
( [ masses ] ) with the appropriate derivatives of the @xmath7-dependent potentials @xmath126 .
all the dependencies on @xmath127 dropped out after performing the momentum trace .
we calculate the free energy @xmath128 given by eq .
( [ sigma ] ) supplemented by the flow eqs ( [ u_casi1 ] ) and ( [ u_casi2 ] ) , which are integrated for a sequence of @xmath7-values .
( [ sigma ] ) indicates that the thermal energy @xmath129 sets the scale for @xmath128 and therefore also for the casimir force .
we plot the difference @xmath130 $ ] versus @xmath7 in the double - logarithmic scale , as exemplified in fig . 4 for @xmath4 and @xmath5 .
the points tend to lie on a straight line for sufficiently large @xmath7 , indicating scaling behavior .
$ ] versus @xmath7 for @xmath4 , @xmath5 . fitting a straight line to the scaling region
yields the slope @xmath131 and the free coefficient @xmath132 .
this translates to the computed casimir force decaying as @xmath133 and the casimir amplitude @xmath134 , which compares well to the exact result @xmath135.@xcite deviation from scaling behavior occurs for large and small @xmath7 , where the condition @xmath136 becomes violated.,width=321 ] by fitting a straight line in the regime @xmath136 , we find the slope @xmath137 and the free coefficient @xmath132 .
this translates to the computed casimir force decaying as @xmath133 .
the critical casimir force amplitude @xmath138 is defined via @xmath139 and the general expectation is that @xmath140 .
the casimir amplitude can be computed from the fit coefficient @xmath141 . in the present case @xmath4 , @xmath5 we find @xmath134 , which compares well to the exact result @xmath135.@xcite the above discussion exemplifies the procedure for the case @xmath4 , @xmath5 , where all results can be compared to exact values . in principle
, the present approximation is least reliable for @xmath4 , but nonetheless our result turns out to agree with the exact one .
the method is obviously not restricted to this case and below we present our results for the casimir force amplitudes @xmath2 for @xmath5 and @xmath16 and varying dimensionality @xmath1 between @xmath142 and @xmath143 . for each considered @xmath1
we plot the difference @xmath130 $ ] vs @xmath7 ( in a logarithmic scale ) . by fitting a straight line
we find the free coefficient @xmath141 together with the error and use eq .
( [ f_c ] ) to extract the casimir amplitude supplemented with the corresponding error . in this subsection
we discuss the ising universality class fixing @xmath5 , and perform the procedure outlined above varying @xmath1 between 2 and 3 . for these values of @xmath1 ,
the obtained values of @xmath144 which govern the decay of the casimir force agree with the expected behavior @xmath140 .
the casimir force amplitude as a function of @xmath1 is plotted in fig .
5 . as function of dimensionality @xmath1 in the ising universality class @xmath5 .
the extra point at @xmath4 ( square ) corresponds to the exact result , and in @xmath6 ( triangle ) to the prediction of mc simulations of ref . .
, width=321 ] as already mentioned , the computed value of @xmath138 in @xmath4 agrees with the exact result @xmath145.@xcite for @xmath6 we obtain @xmath146 .
this may be compared to numerical mc simulations of ref . , which give @xmath147 .
the variation of @xmath148 is much more pronounced within the range @xmath149 $ ] than for @xmath150 , where a constant value is compatible with our results within the errorbars .
the relative error resulting from the fitting procedure outlined in the previous section is about @xmath151 around @xmath4 and increases to @xmath152 around @xmath6
. the accuracy decreases for growing @xmath1 because the quantity @xmath153 computed via eq .
( [ sigma ] ) becomes smaller ( at fixed @xmath7 ) and therefore the maximal value of @xmath7 that can be reliably handled numerically is lowered .
let us also compare the present result in @xmath6 with previous rg studies relying on the @xmath13-expansion .
the two - loop result of ref . retains terms to order @xmath13 and yields @xmath154 . as discussed in ref .
( see also ref . ) , the @xmath13-expansion is however problematic in the present case of periodic boundary conditions , yielding at three - loop level terms nonanalytic in @xmath13 .
the next - to - leading contribution is of the order @xmath155 and leads to @xmath156 corrected to @xmath157 in @xmath6 .
the computation proceeds along the same lines in the present case , where we put @xmath16 .
this yields the extra contribution to the effective potential flow described by eqs ( [ u_casi1 ] ) and ( [ u_casi2 ] ) resulting from the presence of the goldstone mode . as already discussed , the approximation level is too low to access the kosterlitz - thouless physics in @xmath4 and we limit the range of @xmath1 to @xmath158 $ ] .
the casimir force amplitude @xmath159 is plotted as a function of @xmath1 in fig .
6 . as function of dimensionality @xmath1 in the xy universality class @xmath16 .
the extra point in @xmath6 ( triangle ) corresponds to the prediction of mc simulations .
, width=321 ] similarly to the case @xmath5 , we observe a variation in the dependence @xmath159 for @xmath160 , while for @xmath150 our results are consistent with @xmath161 .
we find @xmath162 , which may be compared to the mc value@xcite @xmath163 .
similarly to the case @xmath5 , the mc result is lower and outside our errorbars .
our prediction is again closer to the mc results than the earlier perturbative rg studies , which yield@xcite @xmath164 .
the dependencies depicted in figs 5 , 6 are qualitatively similar , in particular we do not observe anything dramatic happening upon reducing @xmath1 towards @xmath4 in the @xmath165 case ( fig .
it is also worthwhile noting the relatively large errorbars resulting from the fitting procedure illustrated in detail at the beginning of the section .
the errors account only for the numerical inaccuracies .
we have nothing to say here about the errors caused by the approximation made while deriving the flow equations .
as already remarked , the errorbars are wider for @xmath166 than in @xmath110 .
the inaccuracies are larger than one might perhaps expect , because calculation of the casimir amplitude requires finding the free coefficient in the fit straight line given only points far from @xmath167 . for this reason a small tilt variation causes sizable changes in the point of interest , where the fit line intersects the @xmath168-axis .
this results in significantly worse accuracy in determining the free coefficient than the slope . the calculation presented in sec .
iii , iva and ivb relies on the ansatz ( [ lpa_ansatz ] ) , and , as described in sec .
iii , gives inaccurate values of the bulk critical indices for @xmath110 . on the other hand , the value of the casimir amplitude obtained for @xmath4
, @xmath5 compares very well to the exact number .
the quality of the result for @xmath138 in this case does not therefore seem to suffer from the inaccuracies in resolving the bulk critical behavior . here
we present an extension of the ansatz ( [ lpa_ansatz ] ) accounting for the flow of the @xmath169-factor .
this refinement leads to nonzero bulk @xmath99 exponent and much more accurate estimate of the correlation length exponent @xmath94 .
we perform an explicit computation for @xmath4 and @xmath5 , where @xmath99 is largest , and show that the inclusion of @xmath99 has in fact no impact on our estimates of the casimir forces .
we now consider eq .
( [ lpa_ansatz ] ) where the coupling @xmath29 depends on the cutoff scale @xmath34 , therefore making the replacement @xmath170 , where @xmath171 .
the flow equations for the effective potential ( [ u_casi1 ] ) and ( [ u_casi2 ] ) remain valid upon substituting @xmath172 and @xmath170 , respectively .
so does eq .
( [ u_flow ] ) , where the term @xmath173 becomes replaced by @xmath174 , and @xmath99 is defined via @xmath175.@xcite the flow of the @xmath169-factor is extracted along the standard reasoning.@xcite functional differentiation of eq .
( [ wetterich_eq ] ) yields the flow of the inverse propagator @xmath176 $ ] , and the laplacian of the resulting equation at @xmath177 corresponding to the minimum and at zero momentum determines @xmath178 . the truncation level and the way of extracting @xmath178
are fully equivalent to the procedure of refs .
for the flowing @xmath169-factor at @xmath179 we obtain @xmath180 ^ 2 } { [ z_\lambda^l\lambda^2+{u_{\lambda}^l}'(\rho_0)+2\rho_{0 } { u_{\lambda}^l}''(\rho_0)]^5}\times \nonumber\\ \times \big[-2z_\lambda^l\left(z_\lambda^l\lambda^2+{u_{\lambda}^l}'(\rho_0)+2\rho_{0 } { u_{\lambda}^l}''(\rho_0)\right)\lambda^{d-1}\times \nonumber\\ \times \left(1 + 2\mathcal{s}(d)\right)+ \frac{8{z_\lambda^l}^2}{d+1}\lambda^{d+1}\left(1 + 2\mathcal{s}(d+1)\right)\big]\ ; . \;\;\;\;\end{aligned}\ ] ] we introduced @xmath181 where in turn @xmath124 . the corresponding equation for infinite @xmath7 reads @xmath182 ^ 2 } { [ z_\lambda^\infty\lambda^2+{u_{\lambda}^\infty}'(\rho_0)+2\rho_{0 } { u_{\lambda}^\infty}''(\rho_0)]^5 } \times \nonumber \\ \times \big[-2z_\lambda^{\infty}\lambda^d [ z_\lambda^\infty\lambda^2+{u_{\lambda}^\infty}'(\rho_0)+2\rho_{0 } { u_{\lambda}^\infty}''(\rho_0 ) ] + \nonumber \\ \frac{8{z_\lambda^\infty}^2 \lambda^{d+2}}{d+2}\big ] \;\;\;\;\;\;\end{aligned}\ ] ] which is equivalent to an analogous expression in ref . .
the flowing bulk anomalous dimension now follows from @xmath183 . upon tuning the system to bulk criticality by the dichotomy procedure of sec .
ii , the flowing @xmath99 attains a fixed - point value @xmath184 ( see fig . 7 ) . at criticality .
the quantity @xmath185 ultimately departs from the fixed point at @xmath184 since the system is tuned to the critical manifold with a finite accuracy.,width=321 ] as expected,@xcite the present truncation overestimates the value of @xmath99 . for the correlation length exponent
we obtain @xmath186 , close to the correct value @xmath187 . a similar calculation carried out in @xmath6,@xcite ( applying a vertex expansion on top of the derivative expansion ) yielded @xmath99 overestimated by a factor of nearly 2 .
the quality of the results for bulk critical exponents in @xmath6 is comparable to @xmath13 - expansion at second order , but the present approach performs much better than perturbative rg for @xmath4 .
we repeated the computation of the casimir amplitude for the ising universality class in @xmath4 within the present truncation capturing nonzero @xmath99 .
for the considered case @xmath99 is known to take the largest value and the local potential approximation is least reliable for resolving the bulk critical behavior .
nonetheless the obtained points in the scaling region lie practically on top of those plotted in fig . 4 and calculated within the local potential approximation .
we conclude , that including @xmath99 does not influence the asymptotic behavior of the casimir forces .
in this work we have performed a renormalization - group study of casimir forces occurring in the classical @xmath1-dimensional @xmath0 models confined to a slab - like geometry and subject to periodic boundary conditions . in this case
the system preserves translational symmetry and the casimir forces arise solely due to fluctuations .
these are long - ranged and universal at criticality .
our approach relies on a truncation of the one - particle - irreducible variant of functional rg at leading order in the so - called derivative expansion .
we believe this approach complements the mc calculations in that the directly accessible system sizes are by far larger . on the other hand
, it does not suffer from complications occurring in perturbative rg treatments and related to nonanalyticities in the @xmath13-expansion , revealed beyond two loops .
it is also applicable both in @xmath1 close to 2 and 3 , and we scanned a range of @xmath1-values in our calculations .
our results agree with the expectation that the critical casimir force decays with the separation @xmath7 between the walls as @xmath111 and the thermal energy @xmath129 sets the relevant scale .
we have computed the critical casimir force amplitudes @xmath2 for the ising ( @xmath5 ) and xy ( @xmath16 ) universality classes varying dimensionality between 2 and 3 and remaining in the symmetric phase , but very close to the critical point .
our findings in @xmath4 compare well to the exact results . in @xmath6
the computed casimir amplitudes reasonably agree with numerical simulations , they are also significantly closer to mc than the earlier field - theoretic rg calculations .
the level of agreement ( in particular in @xmath4 and @xmath5 ) suggests that the details of the momentum dependencies of the correlation function are not relevant for the magnitude of the casimir forces at criticality .
this may seem somewhat surprising , considering that the presently studied effect arises exclusively due to critical fluctuations .
indeed , our calculation neglects the nonzero anomalous dimension @xmath99 whatsoever and also gives incorrect values of the correlation length exponent @xmath94 in @xmath4 .
still , the results agree well with the exact findings in @xmath4 .
this result suggests considering the casimir amplitudes among the critical properties which are insensitive to the presence of anomalous propagator scaling ( nonzero @xmath99 ) quite alike , for example , the power laws describing the casimir forces decay or the universal shapes of the transition lines in quantum - critical systems . to verify this
, we performed an additional calculation in @xmath4 , @xmath5 capturing the anomalous dimension .
althought this refinement leads to a much better resolution of the bulk critical behavior , it has practically no impact on our results referring to the casimir forces . from a principal point of view
, the present approach is applicable for all the @xmath0 models at arbitrary dimensionality @xmath1 with periodic boundary conditions .
one exception is the kosterlitz - thouless case , which requires going to higher order in derivative expansion . in practice
however , the computation of the casimir forces requires increasing numerical accuracy with growing @xmath1 , and may become problematic already at @xmath188 .
another interesting regime is @xmath189 , where one may again compare to exact results
. it might be interesting to extend the present work to account for boundary conditions explicitly breaking translational symmetry .
this problem is of interest also from the point of view of comparison to experiments performed on helium and fluid mixtures .
the other interesting extensions which should be within the range of the present approach include computations within the ordered phase , where the soft fluctuations arise due to continuous symmetry breaking , and extracting the scaling functions for the casimir force .
h. b. casimir , proc .
wet . , * 51 * , 793 ( 1948 ) .
mostepanienko , n.n .
trunov , _ the casimir effect and its applications _ ( clarendon press , oxford , u.k . ,
casimir effect in critical systems _
( world scientific , singapore , 1994 ) .
m. kardar and r. golestanian , rev .
phys . * 71 * , 1233 ( 1999 ) .
m. bordag , u. mohideen , v.m .
mostepanienko , phys . rep . * 353 * , 1 ( 2001 ) .
a. gambassi , j. phys .
series * 161 * , 012037 ( 2009 ) .
brankov , d.m .
dantchev , and n.s .
tonchev , _ the theory of critical phenomena in finite - size systems - scaling and quantum effects _
( world scientific , singapore , 2000 ) .
m. masuda and m. sasaki , phys .
102 * , 171101 ( 2009 ) .
t. matsuda , phys .
d * 84 * , 023501 ( 2011 ) .
g. l. klimchitskaya , u. mohideen , and v. m. mostepanienko , phys .
d * 86 * , 065025 ( 2012 ) .
e. r. bezerra de mello , v. b. bezerra , h. f. mota , and a. a. saharian , phys .
d * 86 * , 065023 ( 2012 ) .
m. krech , phys .
e * 56 * , 1642 ( 1997 ) .
a. mukhopadhyay and b. m. law , phys .
e * 62 * , 5201 ( 2000 ) .
m. fukuto , y. f. yano , and p. s. pershan , phys . rev
. lett . * 94 * , 135702 ( 2005 ) .
m. krech , j. phys .
matt . * 11 * , r391 ( 1999 ) .
a. gambassi , a. macioek , c. hertlein , u. nellen , l. helden , c. bechinger , and s. dietrich , phys .
e * 80 * , 061143 ( 2009 ) .
r. garcia and m. h. w. chan , phys .
* 88 * , 086101 ( 2002 ) .
d. dantchev , m. krech , and s. dietrich , phys .
lett . * 95 * , 259701 ( 2004 ) .
a. ganshin , s. scheidemantel , r. garcia , and m. h. w. chan , phys .
lett . * 97 * , 075301 ( 2006 ) .
a. macioek , a. gambassi , and s. dietrich , phys .
e * 76 * , 031124 ( 2007 ) .
a. hucht , phys .
lett . * 99 * , 185301 ( 2007 ) .
r. zandi , a. shackell , j. rudnick , m. kardar , and l. p. chayes , phys .
e * 76 * , 030601 ( 2007 ) .
m. hasenbusch , phys . rev .
b * 81 * , 165412 ( 2010 ) .
l. canet , b. delamotte , d. mouhanna , and j. vidal , phys .
b , * 68 * , 064421 ( 2003 ) .
h. ballhausen , j. berges , and c. wetterich , phys .
b * 582 * , 144 ( 2004 ) .
g. v. gersdorff and c. wetterich , phys . rev .
b * 64 * , 054513 ( 2001 )
. for recent reviews of the exact rg with applications in different contexts see e.g. b. delamotte , d. mouhanna , and m. tissier , phys .
b * 69 * , 134413 ( 2004 ) ; j.m .
pawlowski , ann . phys . * 322 * , 2831 ( 2007 ) ; w. metzner , m. salmhofer , c. honerkamp , v. meden , and k. schoenhammer , rev .
phys . * 84 * , 299 ( 2012 ) ; o.j .
rosten , phys . rep . * 511 * , 177 ( 2012 ) . | we consider the classical @xmath0-symmetric models confined in a @xmath1-dimensional slab - like geometry and subject to periodic boundary conditions . applying the one - particle - irreducible variant of functional renormalization group ( rg ) we compute the critical casimir forces acting between the slab boundaries .
the applied truncation of the exact functional rg flow equation retains interaction vertices of arbitrary order .
we evaluate the critical casimir amplitudes @xmath2 for continuously varying dimensionality between two and three and @xmath3 .
our findings are in very good agreement with exact results for @xmath4 and @xmath5 . for @xmath6
our results are closer to monte carlo predictions than earlier field - theoretic rg calculations .
inclusion of the wave function renormalization and the corresponding anomalous dimension in the calculation has negligible impact on the computed casimir forces . |
junctions consisting of a single organic molecule between two metallic leads hold great promise for future nanoscale devices , where their potential applications include switches , transistors , and sensors .
experimentally , it has proved difficult to control their production in an atomistic manner , and so theoretical studies are crucial for a full understanding of their behaviour .
it is known that inelastic effects play an important role in the behaviour of such devices @xcite , but as yet we lack a full understanding of the processes at play that will lead to a complete interpretation of experimental results . in this paper
, we use a model - system nanojunction including electron - vibration coupling @xcite in order to determine , by investigating the whole parameter space , what level of diagrammatic expansion is appropriate to describe the electron - vibration interaction in such junctions . we cover the entire parameter space , thus accounting for all the physical analogues to our model .
the parameters we can vary correspond to the lead - molecule - lead coupling , the electron - vibron coupling strength , and the resonance of the electronic level with the leads electronic states in general , an organic molecule - based nanojunction is unlikely to have its homo or lumo levels in alignment with the equilibrium fermi level of the leads , and so such a nanojunction will be dominated by what we term the off - resonant regime with strong tunneling at low bias .
moreover , coupling between the leads and the central molecule is likely to be relatively weak , in the sense that the corresponding coupling to the leads is much smaller than the corresponding hopping integrals in the leads themselves .
conversely , a system consisting of a nanoconstriction in a gold wire will have electronic levels in the constriction that are close to those of the leads , and a larger coupling between the central region and the leads , closer to the tight - binding hopping parameter of the leads
. this work will distinguish between these regimes and discuss for which physical systems diagrams beyond the self - consistent born approximation ( scba ) become relevant for electron - phonon coupling .
we study this using a full non - equilibrium greens - function ( negf ) technique @xcite which allows us to study all the different transport regimes in the presence of electron - vibron interaction . following the spirit of many - body perturbation theory and feynman diagrammatics
, we include the commonly - used scba diagrams as well as second - order diagrams in terms of the electron - vibron interaction .
a detailed description of the formalism and the numerical implementation of the negf code we have developed is given in ref .
[ ] . in this work we studied the equilibrium and non - equilibrium electronic structures of the nanojunctions in the presence of electron - vibron coupling .
in the present paper , we now give and analyse results for the full non - equilibrium transport properties , namely the non - linear @xmath0 characteristics , the conductance @xmath1 , and especially the iets signal @xmath2 calculated with our negf code .
the paper is structured as follows . in section [ sec : theory ] we summarize the key aspect of our methodology detailed in ref . [ ] .
results for the effects of the second order diagrams on the non - equilibrium non - linear transport properties are presented in section [ sec : results ] .
they are separated into the features we observe for the purely inelastic effects in the iets signal at bias equal to an integer multiple of the vibron energy , and for the inelastic resonant features related to the vibron replica of the electronic resonances .
our conclusions are given in section [ sec : conclusions ] .
in addition , we explain in detail in appendix [ sec : appendix - vertex ] how one of the second - order diagrams acts as a vertex correction to the fock - like electron - vibron diagram .
a fully atomistic description of the non - equilibrium inelastic transport properties we wish to study is , unfortunately , beyond the reach of current _ ab initio _ methods .
instead we use a model system which retains the essential physics of the junction while reducing the calculations to a tractable size .
a full description of both the model and our methodology is given in ref .
[ ] , and so we review only the most salient features here .
we use the single - site single - mode model ( sssm),in which the central molecule of the junction is represented by a single molecular level coupled to a single vibrational mode . we have already used this model and discussed its validity in our previous study on the equilibrium and non - equilibrium electronic structures of such a system coupled to two electron reservoirs @xcite .
the total hamiltonian for the nanojunction is given by @xmath3 in this work , we represent the hamiltonian of the left ( @xmath4 ) and right ( @xmath5 ) leads @xmath6 with a non - interacting tight - binding model with semi - elliptic bands , although in principle it can take any valid form .
the hopping between leads and the central region is given by @xmath7 , where @xmath8 is the hopping integral between the @xmath9 lead and the central region .
the central region contains the electron - vibron interactions .
we choose that an electron couples linearly , via its density , to the displacement of a single vibration mode .
the hamiltonian for the central region in the sssm model is then given by @xmath10 where @xmath11 ( @xmath12 ) creates ( annihilates ) an electron in the molecular level @xmath13 , which is coupled to the vibration mode of energy @xmath14 via the coupling constant @xmath15 .
a detailed analysis of the formalism of the non - equilibrium transport properties from negf and for interacting system is provided in ref .
the current @xmath16 passing through each lead @xmath17 is expressed in terms of two time green s functions@xcite .
it is transformed into frequency representation for the steady - state regime to give : @xmath18 we vary the applied bias by moving the chemical potentials of the left and right leads . with the equilibrium
fermi energies @xmath19 , we introduce a quantity @xmath20 such that @xmath21 and @xmath22 following ref . [ ] . in this way we can create several forms for the potential drop across the junction . by setting @xmath23 , for example , we create an asymmetric drop whereby @xmath24 remains constant while @xmath25 is changed , whereas @xmath26 gives a symmetric potential drop where @xmath25 rises ( lowers ) as @xmath24 lowers ( rises ) by the same amount .
it now remains to construct our non - equilibrium green s functions ( details given in ref . ) .
the retarded and advanced green s functions are calculated using a dyson equation @xmath27 while the greater ( @xmath28 ) and lesser ( @xmath29 ) green s functions are obtained from a quantum kinetic equation with the form @xmath30 where @xmath31 is the _ non - interacting _
green s function of the isolated central region . in this work
we consider both first- and second - order contributions to the electron - vibron coupling .
the first - order diagrams are shown in fig .
[ fig : fockhartree ] , and , if calculated self - consistently , equate to the commonly - used self - consistent born approximation ( scba ) .
we also make use of the two second - order diagrams , those which involve two phonon excitations ( fig .
[ fig : dph - dx ] ) .
the first of these is similar in structure to the @xmath32 skeleton for electron - electron interactions , and consists of a fock - like diagram where the phonon is dressed by a single electron - hole bubble ( hence the appellation dph for dressed - phonon diagram , fig .
[ fig : dph - dx ] left ) .
the second , which we call the double - exchange ( dx ) diagram ( fig .
[ fig : dph - dx ] right ) , includes two phonons simultaneously , with the second being emitted before the first is reabsorbed .
the dx diagram is part of the skeleton diagrams corresponding to vertex corrections .
we use these diagrams to construct expressions for the electron - vibron self - energy @xcite as the ( total or partial ) sum of each diagram : @xmath33 . note that in order to handle numerically sharply peaked functions or strongly discontinuous functions ,
we have found it necessary to include a very small but finite imaginary part in our expression for the bare vibron green s function @xcite .
this also allows us to perform calculations with a smaller number of @xmath34-grid points , as long as our imaginary part @xmath35 in the bare vibron green s function is around two to three times the @xmath34-grid spacing .
we have already discussed in detail the effects of the corresponding extra broadening on the lineshape of the spectral functions and on the values of the linear conductance in ref .
in this section we present the effects of the second - order diagrams on the full non - equilibrium transport properties of the nanojunction in the presence of electron - vibron coupling .
calculations of the green s functions are performed with different levels of approximation for the electron - vibron self - energies ( figs .
[ fig : fockhartree ] and [ fig : dph - dx ] ) .
fully self - consistent calculations using the first - order diagrams are annotated scba , those using one or both second - order diagrams are annotated sc(ba+dx ) , sc(ba+dph ) or sc(ba+dx+dph ) as appropriate .
in addition we have performed non - self - consistent second - order corrections , i.e. by using the scba green s functions to calculate the second - order diagrams , we then determine the new green s functions without full self - consistency .
these calculations are annotated scba+dx or scba+dph .
the inelastic properties of the system are present in the current @xmath0 but are better represented by the second derivative of the current @xmath2 as it is the signal that is directly measured experimentally in the form of the inelastic electron tunneling spectrum ( iets)@xcite
. the iets curves present features , peaks or dips @xcite at biases corresponding to the energy of a specific excitation , in our case to the energy of one or several excitations of the vibration mode .
the peak feature is commonly associated with the opening of a new inelastic channel for the conductance of nanojunctions in the off - resonant regime , i.e. when the electronic level @xmath13 is sufficiently far from the equilibrium fermi level . in the case of the resonant transport regime ( when @xmath13 is close to @xmath36 ) , a dip feature is obtained in the iets .
it is associated with electron - vibron backscattering effects and a decrease in the conductance at the threshold bias .
furthermore , being the derivative of the conductance , the iets curves also present features at biases corresponding to peaks in the conductance .
we have found @xcite that in order to get a better aspect ratio for the iets features corresponding to vibron excitations , it is more convenient to normalize the iets curves by the dynamical conductance , i.e. @xmath37/[di / dv]=d / dv \ln g(v)$ ] as in refs . .
we divide our results into two sections ; the first for purely inelastic features , and the second for inelastic features associated with the electron resonance effects .
the first category corresponds to features observed in the iets signal at bias equal to an integer multiple of the vibron energy @xmath38 , i.e. a tunneling electron excites @xmath39 vibrons .
the second category corresponds to inelastic resonant tunneling via the vibron replica associated with the main electronic resonance at @xmath40 , and hence are observed in the iets signal for biases @xmath41 .
we will see below that the second - order diagrams have different effects on the iets features depending upon their correspondence to one of these two categories .
we consider first of all the off - resonant regime . in this limit
the iets features associated with inelastic resonant tunneling are sufficiently far ( or sufficiently small for the higher vibron replica ) from the inelastic feature at @xmath42 .
we can thus avoid a superposition of the two different kinds of features .
for an asymmetric potential drop ( @xmath43 , left panel ) and a symmetric potential drop ( @xmath44 , right panel ) .
results are shown for three different lead - molecule coupling parameters @xmath45 ( black , red , and blue lines ) , and for two different electron - vibron couplings ( medium @xmath46 , solid lines and triangles ; strong @xmath47 , dashed lines and circles ) .
lines represent calculations at the scba level , symbols sc(ba+dph ) , from which we can see there is no distinguishable contribution to the inelastic peak from dph for any parameter set .
the other parameters are @xmath48 . ]
results are presented in fig .
[ fig : w0peak ] for different electron - vibron couplings and for both symmetric and asymmetric potential drops .
we note that , as would be expected intuitively , the size of the feature increases with the electron - phonon coupling @xmath15in fact the height with respect to the baseline is proportional to @xmath49 .
as @xmath8 is increased , the iets signal decreases in overall magnitude , although the feature at @xmath14 itself remains clear ( fig .
[ fig : w0peak ] inset ) . in the off - resonant regime
, we do not find any difference to the curves when the second - order dressed - phonon ( dph ) diagram is included .
possible reasons for this are discussed in section [ sec : electron - resonance ] .
we also note that the results are dependent on the symmetry of the potential drops at the left and right contact . as shown in the separate panels of fig .
[ fig : w0peak ] , the curves for symmetric potential drops have a much lower baseline than for the asymmetric potential drops , although the inelastic features have the same lineshape , position and magnitude ( see figs . [
fig : w0peak ] , [ fig : scba+dx ] and [ fig : iets_badx_andfit ] ) .
this is because ( for positive bias ) in the asymmetric case , @xmath25 is rising bias while @xmath24 is kept fixed at its equilibrium value .
thus @xmath25 is approaching the electron resonance twice as fast as in the symmetric case , where @xmath25 rises at the same rate as @xmath24 falls .
the symmetric case therefore contains much less effect from the tail of the main electron resonance , and allows us to further isolate the purely inelastic part of the iets signal .
we now consider the effect of the double - exchange diagram ( fig .
[ fig : dph - dx ] right ) on the inelastic peak at @xmath14 .
calculation of @xmath50 is extremely computationally intensive , as it can be reduced neither to a simple convolution product , nor even to a simple double - convolution product @xcite .
calculations for the integration in the energy representation of the green s functions and the self - energies scale as the cube of the number @xmath51 of points in the energy - grid @xcite .
moreover , as the finite imaginary part @xmath35 we have included in the vibron green s function causes extra broadening , if taken too large it washes out much of the effect of the dx diagram . in order to decrease @xmath35
we need to increase the number of grid points , and thus , even for our minimal model , a fully self - consistent calculation of the double - exchange diagram becomes intractable with @xmath52 .
hence for the dx calculations , we usually work with @xmath53 , giving @xmath54 for the total spectral width considered in our calculations . .
results for scba and sc(ba+dph ) are virtually identical .
however , including the dx diagram , either self - consistently ( solid red line ) or non - self - consistently ( red diamonds ) raises the amplitude of the peak .
this effect can be approximated by an scba calculation done with a larger @xmath15 ( black triangles ) , indicating that dx generates an effective renormalization of @xmath15 .
the other parameters are @xmath55 . ]
vertex . ] vertex . ]
we work around this by calculating the effects of the dx diagram both in a self - consistent manner sc(ba+dx ) or non - self - consistenly as a second - order correction to the scba result .
the results of this are plotted in fig .
[ fig : scba+dx ] .
we see that the effect of the dx diagram is to both increase the height of the feature and to raise its baseline . for the set of parameters
shown the self - consistency in the dx diagram calculations is not crucial .
we see that the effect of dx can also be reproduced with a scba calculation in which we increase the value of @xmath15 .
therefore the dx diagram has the effect of renormalizing @xmath15 as it is part of the skeleton family of vertex correction , as shown in fig .
[ fig : dx - feynman - mod ] .
we can thus approximate the effect of the dx diagram in the iets with a fock - like diagram with one renormalized vertex @xmath56 .
the amplitude of the peak at @xmath14 , instead of varying as @xmath49 , therefore now depends on @xmath57 , and so we can define an _ effective _ electron - vibron coupling constant @xmath58 , with @xmath59 .
this allows us to make a more quantitative analysis of @xmath60 by fitting the scba+dx curve to an scba curve with electron - vibron coupling @xmath61 , as shown in fig .
[ fig : scba+dx ] .
furthermore , we can study how @xmath56 and @xmath15 are correlated by performing a series of calculations for different values of the parameters @xmath15 and @xmath14 , and then fitting the scba+dx results to those from scba calculations .
the results of this are shown in fig .
[ fig : gamma0correlation ] , from which we can see that , to a first approximation , the dx diagram consistently raises the effective electron - vibron interaction by approximately 3% within the range of parameters we used . as a function of the nominal coupling parameter @xmath15 , from a fit of several sets of data comparing scba+dx curves to scba ( as in fig .
[ fig : scba+dx ] ) .
the data for all three values of @xmath14 lie on a straight line with slope 1.03 ( the straight line with slope 1 is shown for comparison ) , implying that the vertex correction increases the effective electron - vibron coupling by 3% . ]
however , we would like to point out that although the apparent effects of the dx diagram on the iets signal is to renormalize the coupling constant @xmath15 , the reality is much more subtle . in appendix
[ sec : appendix - vertex ] we discuss in detail the renormalization effect of the dx diagram in terms of vertex corrections , and we show that such vertex corrections do not simply correspond to a mapping of the sssm hamiltonian onto a similar hamiltonian with a static renormalization of the electron - vibron coupling constant , i.e. @xmath62 . rather , the vertex correction actually generates a dynamical renormalization of the electron - vibron coupling constant , i.e. @xmath63 .
this can be seen more clearly by considering scba calculations , for different values of the electron - vibron coupling parameter , and checking which of such calculations correspond the best to a sc(ba+dx ) calculation .
the result is shown in fig . [
fig : iets_badx_andfit ] .
although the difference between the best scba fits and the sc(ba+dx ) are not large , it is quite clear that a renormalized scba calculation does not provide exactly the same lineshape as a full sc(ba+dx ) calculation for all the range of biases @xmath64 around the inelastic peak at @xmath14 . showing different scba fits for the renormalized electron - vibron coupling .
the scba ( dashed line ) and sc(ba+dx ) ( solid red line ) are shown for @xmath65 , together with scba calculations for @xmath66 and @xmath67 .
although these provide a good approximation , neither gives an exact fit to the lineshape of the sc(ba+dx ) curve .
the other parameters are @xmath68 . ]
finally we expect to see a peak feature at @xmath69 in the iets for the off - resonant transport regime .
this peak feature is the two - vibration excitation equivalent of the feature observed at @xmath70 . since this a higher - order process
, the amplitude of the feature should be @xmath49 times smaller than the feature at @xmath71 .
this feature has a rather small amplitude for all the electron - vibron coupling constants we have considered in this work since @xmath72 .
an example of a close - up of the iets feature around @xmath69 is given in figure [ fig : offres_ietssidebandpeak ] .
we find that the amplitude of the peak with respect to the linear baseline is indeed approximately @xmath73 ( one order of magnitude ) smaller than the corresponding amplitude of the peak at @xmath74 ( shown in fig .
[ fig : scba+dx ] ) .
once more we find that the effects of the dph diagram are negligible for this part of the iets signal . in fig .
[ fig : offres_ietssidebandpeak ] we also include the results of a scba calculation for a larger coupling @xmath75 which mimics the renormalization effects of the dx diagram as discussion above . for different levels of approximation
: the scba ( solid black line ) and sc(ba+dph ) ( red circles ) are virtually identical , as for the feature at @xmath14 . an scba calculation approximating the effect of dx
is also included ( dashed line ) .
a straight baseline is included for clarity .
the other parameters are @xmath76 . ]
the main electron - resonance peak occurs at the polaron - shifted value of @xmath77 , and consists of a peak - dip feature in the iets , as it corresponds to a resonant peak feature in the conductance . with no electron - phonon coupling ( @xmath78 )
, the iets curve has no features other than the one corresponding to the resonant transmission in the conductance at @xmath79 .
once the electron - phonon coupling is turned on , phonon side - band peaks emerge in the spectral function at energies @xmath80 .
the @xmath81 features correspond to phonon emission ( vibration excitation ) by an electron , while the @xmath82 features correspond to phonon - emission by a hole . in the iets , they appear as peak - dip features ( derivatives of an inelastic resonant peak in the conductance ) with amplitude decreasing as the bias is further increased . at lower biases however , there are no features at @xmath83 at the scba level except for very low values of the coupling to the leads ( @xmath84 , fig .
[ fig : resonance - peaks ] top inset ) . including the dph diagram
, however , introduces a small peak - dip features at @xmath85 just below the main peak - dip feature at @xmath86 in the iets , as shown in fig .
[ fig : resonance - peaks ] ) .
the dph diagram also has a strong influence on the lineshape of the other phonon side - band peaks above @xmath87 , increasingly so as @xmath13 is brought within range of the equilibrium chemical potentials ( i.e. @xmath88 in fig .
[ fig : resonance - peaks ] bottom ) . .
with no electron - vibron coupling ( solid black line ) , there is a single feature at @xmath89 . with electron - vibron coupling ( red and blue dashed lines ) ,
the main peak moves to @xmath77 and side - band peaks appear .
the inset shows a tiny feature at @xmath90 that only occurs for @xmath91 in scba calculations but reappears for all @xmath92 once dph is included .
bottom : the inelastic resonant features for various @xmath13 at @xmath93 for both scba ( lines ) and sc(ba+dph ) ( symbols ) calculations , showing the increasing influence of dph in the resonant transport regime @xmath94 .
the other parameters are : @xmath95 .
, title="fig : " ] + . with no electron - vibron coupling ( solid black line )
, there is a single feature at @xmath89 . with electron - vibron coupling ( red and blue dashed lines ) ,
the main peak moves to @xmath77 and side - band peaks appear .
the inset shows a tiny feature at @xmath90 that only occurs for @xmath91 in scba calculations but reappears for all @xmath92 once dph is included .
bottom : the inelastic resonant features for various @xmath13 at @xmath93 for both scba ( lines ) and sc(ba+dph ) ( symbols ) calculations , showing the increasing influence of dph in the resonant transport regime @xmath94 .
the other parameters are : @xmath95 .
, title="fig : " ] .
calculations performed for different approximations for the electron - vibron self - energy : ( left panel ) scba ( black solid line ) , scba plus second order dph correction ( dashed red line ) , sc(ba+dph ) ( dot - dashed blue line ) ; ( right panel ) : scba ( solid black line ) , sc(ba+dx ) ( dashed red line ) and sc(ba+dx+dph ) ( dot - dashed blue line ) .
the other parameters are @xmath96 . ]
we now consider the combined effects of the dph diagram as well as of the dx diagram on the specific case of the iets feature at @xmath97 .
this is shown in figure [ fig : electronres_sidebandpeak ] .
the dph diagram increases strongly the peak - dip feature ( at @xmath98 ) obtained from scba calculations at medium / strong electron - vibron coupling .
note here the importance of the self - consistency in the calculations : the second - order dph diagram calculated as a second - order correction to scba ( fig .
[ fig : electronres_sidebandpeak ] left ) gives a completely wrong feature in the iets .
interestingly , the self - consistent calculation with the dx diagram seems to give a similar feature to that observed in the scba calculations , but slightly shifted towards lower bias ( fig .
[ fig : electronres_sidebandpeak ] right ) .
this is completely consistent with the renormalization effects of the electron - vibron coupling by the dx diagram as discussed in the previous section . indeed
, the dx diagram renormalizes the coupling @xmath15 towards a higher value @xmath99 .
consequently the renormalization of the molecular level by @xmath100 is more important than for scba calculations , and thus the feature is moved towards lower bias .
the calculations performed with both dx and dph diagrams ( fig . [ fig : electronres_sidebandpeak ] right ) generate a hybrid feature in the iets in comparison to individual calculations with the second - order diagrams .
however the new iets is not simply obtained by a linear superposition of the individual effects of the dph and dx diagrams .
it might at first seem strange that the dph self - energy is negligble at @xmath101 , where one might expect it to be influential , but that it has a significant effect at biases @xmath102 . while this is to some extent related to the strength of the electron - vibron coupling
, there is another , more important , underlying cause .
the dph diagram involves an electron - hole bubble , and so for this diagram to become relevant , simultaneous electron and hole states must be available .
this is not the case when the spectral functions of the coupled electron - vibron system are mostly empty or mostly filled .
when the bias is significantly low and both fermi levels @xmath103 are below the electron resonance level , these excitations are inaccessible and so there is no effects from the dph diagram .
once the bias is increased to within range of @xmath104 , or @xmath105 however , these electron - hole states become accessible and the dph diagram becomes influential ( unless the spectral function is mostly filled ) .
this is also borne out by the increasing contribution from dph to the lineshape of the phonon side - band peaks as the electron level @xmath13is decreased , moving dph s sphere of influence to lower and lower biases ( see lower panel in fig .
[ fig : resonance - peaks ] ) . having examined the role of the second - order diagrams in detail for characteristic selected sets of parameters , we now show results across the entire parameter range . in order to present this in a concise manner , we have compiled maps of our iets results comparing scba calculations to those with sc(ba+dph ) indicating in which regimes the dph self - energy has a significant effect . for the off - resonant regime ( fig .
[ fig : maps ] left ) , we can see that the greatest effect of dph is apparent at higher bias ( i.e. approaching the electron resonance @xmath13 ) and when the lead - molecule - lead coupling is small . for the resonant case
, however , we can see that the dph self - energy only gives a non - negligble contribution in the region of parameter space where the coupling to the leads is small , and at low bias .
this has potential implications for real molecular junctions .
consider a junction which has its dominant molecular levels far from the equilibrium fermi levels of the leads . at sufficiently high bias
, the dph self - energy will have a significant contribution to the inelastic spectra .
this could occur in the case of a junction formed from an organic molecule if the electronic level of the molecule is within range of the intended operational bias of the junction .
if however the dominant molecular electronic level is close to the leads fermi levels , and the coupling to the leads is large ( as would be the case , for example , in a gold nanoconstriction ) then dph will not give a significant contribution at any applied bias . .
it gives a near - zero result ( blue and green areas ) when the two calculations give the same spectrum , and a positive number ( red areas , i.e. bottom right corner for scba and bottom left corner for sc(ba+dph ) ) where there is a substantial difference between the two .
the other parameters are @xmath106 ( off - resonant / resonant ) , @xmath107.,scaledwidth=49.0% ] .
it gives a near - zero result ( blue and green areas ) when the two calculations give the same spectrum , and a positive number ( red areas , i.e. bottom right corner for scba and bottom left corner for sc(ba+dph ) ) where there is a substantial difference between the two .
the other parameters are @xmath106 ( off - resonant / resonant ) , @xmath107.,scaledwidth=85.0% ]
by using the non - equilibrium green s functions technique , we have studied the effect of electron - vibron interaction on the inelastic transport properties of single - molecule nanojunctions for a model system .
we have included not only the first - order diagrams ( ba ) but also the second - order diagrams ( double - exchange dx and dressed phonon dph diagrams ) for the electron - vibration interaction .
we have calculated the inelastic electron tunneling spectrum ( iets ) across the full range of parameters available to our model .
the effects of the second - order dx and dph diagrams are different and affect different features of the iets signal .
the effects of these diagrams are generally less visible in the integrated quantities , such as the current or the derivated iets signal , than in the spectral functions @xcite .
however their effects are non - negligible , and are important for the full understanding of the spectroscopic information conveyed by the iets signal .
the effect of the dressed - phonon ( dph ) diagram is more important in the bias regions where one of the leads chemical potentials begins to impinge upon the electron resonance or one of its vibron replica ( i.e. for resonant inelastic features ) .
its effect is reduced both by increasing the lead - molecule - lead coupling and/or reducing the electron - vibron interaction .
the renormalization of the vibron propagator ( dph ) has been shown to be strongly dependent on the self - consistency of the calculations .
it would be interesting now to study the effects of the full series of the electron - hole bubble on the renormalised vibron propagator ( i.e. full @xmath32-like diagram ) .
the double - exchange diagram ( dx ) affects all the features in the iets signal ( i.e. resonant inelastic features and purely inelastic features at @xmath108 ) .
the corrections are small in the weak - to - medium electron - vibron coupling because they are of the order of @xmath109 .
however we have shown , numerically and analytically , that the effect of dx is similar to a dynamical renormalization of one vertex in the fock - like diagram .
more interestingly , the complex form of the non - equilibrium dynamical renormalized electron - vibron coupling @xmath110 we have derived analytically can be adequately replaced in our iets calculations by a single static renormalized parameter @xmath61 .
this important result leads us to believe that the second - order dx calculations , which are extremely costly in computing time even for our model system , can be incorporated in calculations for realistic systems by an appropriate renormalization of the vertex in a low - computing - cost scba calculation .
in appendix a of ref . [ ] we have given all the details for the derivations of the first- and second - order electron - vibron self - energies . in this section
we show how the second - order double - exchange ( dx ) diagram can be recast in an effective first - order fock - like diagram with a renormalized vertex .
we recall that for the sssm model , the fock and dx self - energies defined on the keldysh contour @xmath111 are given by @xmath112 and @xmath113 the dx self - energy can be rewritten as an effective fock - like diagram after introducing a renormalized electron - vibron coupling parameter @xmath114 and the vertex function @xmath115 : @xmath116 with @xmath117 and where the vertex function is given by @xmath118 the above expression for the vertex function is compatible with the second - order expansion of the electron - vibron interaction .
a generalization of the vertex function ( see fock - like diagram in figure [ fig : dx - feynman - mod ] ) to all orders of the interaction is possible , though beyond the scope of the present paper .
note that at the lowest order , the renormalized electron - vibron coupling parameter would be given by @xmath119 with @xmath120 .
hence eq.([eq : dxse_onck ] ) would simply be transformed eq.([eq : fockse_onck ] ) as expected . using the rules of analytical continuation on the real - time branches given in appendix a of paper ref .
[ ] , we find the different components of the self - energies . then after taking the fourier transform of the different quantities in the steady - state limit , i.e. @xmath121 , we find the following expression for the different components of the energy - dependent self - energies : ( the index @xmath125 labels the branch of the keldysh time - loop contour @xmath111 and are related to the usual convention : time - ordered ( @xmath126 ) , anti time - ordered ( @xmath127 ) , greater ( @xmath128 ) and lesser ( @xmath129 ) components . ) because all the quantities are originally defined on @xmath111 and because the vertex function is a 3-point ( 3 times ) function , the non - equilibrium dynamical renormalized electron - vibron coupling is a complex function of three indices @xmath130 and of two energy variables .
such a dynamical renormalization ( including non - equilibrium conditions ) is much more complicated than a simple static renormalization of the electron - vibron constant coupling @xmath131 . in our analysis of the iets signal in the off - resonant transport regime
, we try to keep the interpretation of the results as simple as possible , and we show that the renormalization of the iets signal due to the dx diagram can be fairly well approximated by a simple static renormalization of the coupling constant @xmath15 for applied bias around the vibration frequency @xmath132 . | we study the effect of electron - vibron interactions on the inelastic transport properties of single - molecule nanojunctions .
we use the non - equilibrium green s functions technique and a model hamiltonian to calculate the effects of second - order diagrams ( double - exchange dx and dressed - phonon dph diagrams ) on the electron - vibration interaction and consider their effects across the full range of parameter space .
the dx diagram , corresponding to a vertex correction , introduces an effective dynamical renormalization of the electron - vibron coupling in both the purely inelastic and the inelastic - resonant features of the iets .
the purely inelastic features correspond to an applied bias around the energy of a vibron , while the inelastic - resonant features correspond to peaks ( resonance ) in the conductance .
the dph diagram affects only the inelastic resonant features .
we also discuss the circumstances in which the second - order diagrams may be approximated in the study of more complex model systems . |
the effective potential plays a crucial role in determining the nature of the vacuum in a weakly coupled field theory , as was emphasized in the classic paper of coleman and weinberg @xcite .
calculation of this object by summing infinite series of feynman diagrams at zero momentum is an onerous task , especially when several interactions are present which complicate the combinatorial factors that multiply each diagram .
jackiw has succeeded in representing each loop order containing an infinite set of conventional feynman diagrams by finite number of diagrams using his algebraic method which can be formally extended to the arbitrary higher loop order @xcite .
self - interacting scalar field theory of @xmath1 model is one of the best analyzed field theories .
various renormalization group functions of @xmath0 @xmath1 theory are now available up to the five - loop order @xcite . however , the three - loop effective potential of the same theory is less progressive .
for renormalization of the two - point one - particle - irreducible ( 1pi ) green s function , @xmath3 , and the four - point 1pi green s function , @xmath4 , of the @xmath0 @xmath1 theory only one type of propagator is involved in loop integrations , but for calculation of the effective potential two kinds of propagators due to two different induced mass - like terms are involved in the loop integrations .
the two - loop effective potential for the massive @xmath0 @xmath1 theory in four dimensions has been calculated by ford and jones @xcite , and for the massless @xmath0 @xmath1 theory by jackiw @xcite . for the single - component massive @xmath1 theory in four dimensions , the two - loop effective potential has been calculated in ref .
@xcite , and the three - loop effective potential in ref .
@xcite .
recently all the genuine three - loop integrals genuine in the sense that they can not be factorized into lower - loop integrals appearing in the three - loop effective potential calculation of the massless @xmath0 @xmath1 theory has been calculated @xcite .
these integrals carry two kinds of propagator lines , a - type line and b - type line .
the mass parameter of the b - type line is one - third as large as that of the a - type line .
the purpose of this paper is to calculate the three - loop effective potential of the massless @xmath0 @xmath1 theory in four dimensions of space - time in the dimensional regularization scheme @xcite . in sec .
ii the standard procedeure for the renormalization is presented .
the feynman rules for the effective potential suitable for the jackiw s prescription are given and all the feynman diagrams for the effective potential up to three - loop order are constructed .
the values of all relevant integrals are listed in the appendices a and b. sec .
iii is devoted to concluding remarks .
the lagrangian for a theory of @xmath5 spinless fields @xmath6 , with an o(@xmath5)-invariant interaction is given as ( _ a(x))&=&1+z2__a^_a -m^2+m^22 ^ 2-+4!^4 , [ lg ] where the quadratic and quartic expressions are @xmath7 and @xmath8 .
the quantities @xmath6 , @xmath9 , and @xmath10 are the renormalized field , the renormalized mass , and the renormalized coupling constant respectively , whereas @xmath11 , @xmath12 , and @xmath13 are corresponding ( infinite ) counterterm constants .
these are related to the usual bare quantities as follows : m_0 ^ 2=m^2+m^21+z , _0=+(1+z)^2 , _ 0 ^ 2=(1+z)^2 .
we will confine ourselves to the massless theory ( @xmath14 ) .
the effective potential is most suitbly defined generically , when the effective action @xmath15)$ ] , being the generating functional of the one - particle - irreducible ( 1pi ) green s functions ( @xmath16 ) , is expressed in the following local form ( the so - called derivative expansion ) : & = & d^4x , [ loc ] where @xmath17 is the vacuum expectation value of the field operator @xmath18 in the presence of an external source . by setting @xmath17 in @xmath19 to be a constant field @xmath20
, we obtain the effective potential @xmath21 v_eff()(_cl(x))|__cl(x)=.[ve ] following the field - shift method of jackiw @xcite for the calculation of the effective potential , we first obtain the shifted lagrangian with the constant field configuration @xmath22 ( _ a;_a(x))&=&1+z2__a^_a -12_a_b + & -&+6_a_a^2-+4!^4 , [ slg ] where @xmath23 .
our perturbation theory differs from the bare or renormalized perturbation theory @xcite ; rather a mixed one between both of them .
the vertices for the @xmath24-quadratic term with strength @xmath11 and @xmath12 in the usual ` renormalized ' perturbation theory are transferred to the propagator line in our treatment .
thus our perturbation theory follows the one used by kastening @xcite in spirit .
the feynman rules of this shifted lagrangian are ps .
( 0.8mm)(3,3,43,3)(23,-4 ) ( 3,-4)(@xmath25)(43,-4)(@xmath26)&= & ps . ( 0.8mm)(3,3,43,3)(23,-4)(a ) _ a_b^2 + ps .
( 0.8mm)(3,3,43,3)(23,-4)(b ) + & = & i(1+z)k^2-m^2-(+)^2/2 _ a_b^2 + & & + i(1+z)k^2-m^2-(+)^2/6 , + ( -1,0,1,15 ) ( 120,210)(-1,0,1,15 ) ( 240,330)(-1,0,1,15 ) ( -14,-13)(@xmath25)(17,-13)(@xmath26)(0 , 21)(@xmath27 ) & = & 13(_ab_c+_bc_a+_ca_b ) , + ( 45,135)(-13,1,17,3 ) ( 45,135)(1,-13,3,17 ) ( -14,-14)(@xmath25)(16,21)(@xmath28 ) ( -14 , 19)(@xmath27)(16,-14)(@xmath26 ) & = & 13(_ab_cd+_ac_bd+_ad_bc ) , [ jf ] where the propagators without group indices @xmath25 and @xmath26 , but with small capital letters a and b are given , as self - evident in the above expression , by ps .
( 0.8mm)(3,3,43,3)(23,-4)(a ) & = & i(1+z)k^2-m^2-(+)^2/2 , + ps .
( 0.8mm)(3,3,43,3)(23,-4)(b ) & = & i(1+z)k^2-m^2-(+)^2/6.[pab ] the rules for vertices without group indices can also be defined : ( 45,135)(-13,1,17,3 ) ( 45,135)(1,-13,3,17 ) = -i(+ ) , ( -1,0,1,15 ) ( 120,210)(-1,0,1,15 ) ( 240,330)(-1,0,1,15 ) = -i(+).[vert ] the propagators and vertices without the group indices , eqs .
( [ pab ] ) and ( [ vert ] ) , will appear in the final formal expression when the group indices are contracted out . without introducing any new loop - expansion parameter , which is eventually set to be unity
, we will use @xmath29 as a loop - counting parameter @xcite .
this is the reason why we have kept all the traces of @xmath29 s in the feynman rules , eqs .
( [ jf ] ) , ( [ pab ] ) , and ( [ vert ] ) , in spite of our employment of the usual `` god - given '' units , @xmath30 .
in addition to the above feynman rules , eq .
( [ jf ] ) , which are used in constructing two- and higher - loop vacuum diagrams , we need another set of rules solely for one - loop vacuum diagrams which are dealt with separately in jackiw s prescription and are essentially the same as those of coleman and weinberg @xcite from the outset : ( 3,4.3,43,4.3)(3,1.7,43,1.7)(23,-4)(a ) & = & -(1 -(+)^2/2(1+z)k^2-m^2 ) , + ( 3,4.3,43,4.3)(3,1.7,43,1.7)(23,-4)(b ) & = & -(1 -(+)^2/6(1+z)k^2-m^2 ) . [ one ] using the rules , eqs .
( [ jf ] ) and ( [ one ] ) , and including the terms of zero - loop order , we arrive at the formal expression of the effective potential up to the three - loop order ( see fig . 1 to fig .
9 ) : v_eff(_a)&=&+ + + + & + & + + + + + .[onn ] we have revealed both symmetry number factor and group factor ( the second factor when two factors appear simultaneously ) explicitly in front of each diagram in fig . 1 to fig .
9 . for the purpose of renormalization
we first expand the counterterm constants in power series , beginning with order @xmath29 : m^2&=&m_1 ^ 2+^2m_2 ^ 2 + , + &=&_1+^2_2 + , + z&=&z_1+^2z_2 + . as is well known in our original @xmath1 theory of eq .
( [ lg ] ) @xmath31 vanishes @xcite .
thus the field renormalization counterterm appears in the three - loop calculation for the first time . in the first square - bracket term on the right - hand side of eq .
( [ onn ] ) , we need @xmath32- , @xmath33- , @xmath34- , and @xmath35-order terms in its expansion . in calculating [ diag .
1 ] , [ diag . 2 ] , and [ diag . 3 ]
terms , the careful @xmath29 expansions are needed : @xmath33- , @xmath34- , and @xmath35-order terms are needed from [ diag . 1 ] and @xmath34- and @xmath35-order terms from [ diag . 2 ] and [ diag .
3 ] . in the appendix b ,
the details of the counterterm integrals are described . in the remaining terms , [ diag .
4 ] to [ diag . 9 ] , the lowest order in @xmath29 is @xmath35 , which is already of the desired order .
thus with a simple replacement of every @xmath36 in eqs .
( [ 3 t ] ) and ( [ 3 g ] ) with @xmath37 we readily evaluate the three - loop diagrams in fig . 4 to fig
. 9 . in what follows we will use the following notation for the effective potential up to the @xmath38-loop order : v_eff^[l](_a)=_i=0^l ^i v_eff^(i)(_a ) , and use @xmath39 for @xmath40 .
the zero - loop part of the effective potential is given as v_eff^(0)(_a)=4!^4 .
we first remove all the @xmath2 poles in the subsequent contributions to the effective potential , @xmath41 ( @xmath42=1,2,3 ) .
the one - loop part of the effective potential is readily obtained as v_eff^(1)(_a)&=&m_1 ^ 22 ^ 2+_14 ! ^4 -^2
^ 4(4)^2 + ^2 ^ 4(4)^2.[v1p ] the @xmath2 poles in this equation are readily cancelled out by choosing the counterterm constants @xmath43 and @xmath44 as follows : m_1 ^ 2&=&(4)^2a_1 , + _1&=&^2(4)^2 , where @xmath45 and @xmath46 are unspecified but finite constants at this stage .
one may put @xmath45 ( and @xmath47 , @xmath48 below ) to be zero from the beginning because the theory is massless . in our dimensional regularization scheme the pole part of @xmath43 vanishes , but this is not true in the cutoff regularization method .
the two - loop part of the effective potential is obtained as v_eff^(2)(_a)&=&m_2 ^ 22 ^ 2+_24!^4 -a_1^2 ^ 2(4)^4-^3 ^ 4 ( 4)^4 ^ 2 + & + & ^3 ^ 4(4)^4 -b_1^3 ^ 4(4)^4 + & + & a_1^2 ^ 2(4)^4 + & + & b_1^3 ^ 4(4)^4 + & + & ^3 ^ 4(4)^4 .[v2p ] notice that the so - called `` dangerous '' pole terms such as @xmath49 $ ] , @xmath50 in the above equation , which can not be removed by terms of counterterm constants [ @xmath51 and @xmath52 , have been completely cancelled out among each other .
this serves as a strong check of the correctness of the calculation at this stage .
the counterterm constants @xmath53 and @xmath54 are determined as m_2 ^ 2&=&^2(4)^4 , + _2&=&^3(4)^4 , where @xmath47 and @xmath55 are also unspecified but finite constants . in the minimal subtraction scheme ( ms scheme ) , the finite parts of the counterterm constants ( @xmath45 , @xmath46 , @xmath47 , and @xmath55 ) are taken to be zero . in our massless theory , however , we encounter the infrared singularity in the defining condition for a coupling constant . to avoid this difficulty we follow coleman and weinberg @xcite and require |_=0=0 , d^4 v_effd ^4|_=m = ,[rc ] where @xmath56 is a renormalization scale .
then constants @xmath45 , @xmath46 , @xmath47 , and @xmath55 are determined as a_1&=&a_2=0 , + b_1&=&-4 - 32 - 32(/2 4)-\{49 + 6 + 16 ( /64 ) } , + b_2&=&1394 - 3a+15 + 94 ^ 2 + ( 15 + 92)(/24 ) + 94 ^ 2(/24 ) + & + & \{20227 - 29a+299+^22 -1493 - 23 + & + & ( 299 + -123 ) ( /24)+12 ^2(/24 ) } + & + & ^2\{113324 + 427+^236 + ( 427 + 18)(/64 ) + 136 ^ 2(/64)}. [ a12b12 ] putting these constants into the finite parts of @xmath57 and @xmath58 , we eventually obtain the finite effective potential , @xmath59}$ ] , which satisfies the renormalization conditions , eq .
( [ rc ] ) : v_eff^[2](_a)&= & + ^2 ^ 4(4)^2 + & + & ^2^3 ^ 4(4)^4 .
having illustrated our strategy , we continue with three - loop structure of the model , which is appreciably more complicated than that of previous subsection ii.a . in the @xmath35-order calculation
we need another counterterm constant @xmath60 in addition to the counterterm constants , @xmath43 , @xmath61 , @xmath53 , @xmath62 , ( @xmath63 , and @xmath64 ) @xcite . instead of determining @xmath60 from the renormalization of @xmath65 in eq .
( [ loc ] ) @xcite , we can obtain it from the renormalization of @xmath66 : z_2=^2(4)^4 , ( c_2 ) . using the integrals in appendices a and b , we have the rather long expression for the three - loop part of the effective potential
. thus we will separate it into two parts , the counterterm plus pole part and the finite part .
the counterterm plus pole part is calculated as [ v_eff^(3)]_ct+pl&= & m^2_32 ^ 2+_34!^4 -a_1^3 ^ 2(4)^6
^ 2 + & + & ^3 ^ 2(4)^6 + & -&^4 ^ 4(4)^6 ^ 3 + & + & ^4 ^ 4(4)^6 ^ 2-b_1^4 ^ 4(4)^6 ^ 2 + & -&^4 ^ 4(4)^6 + & + & ^4 ^ 4(4)^6 .
notice that the complete cancellation of all dangerous pole terms [ @xmath67 , @xmath68 ; @xmath69 , @xmath70 , @xmath71 has taken place here too .
remaining harmless pole terms are eliminated by choosing the counterterm constants as follows : m_3 ^ 2&=&^3(4)^6 + _3&=&^4 ^ 4(4)^6,[dl3 ] where @xmath48 and @xmath72 are arbitrary but finite constants as before . when taken in the ms scheme , @xmath73 in eq .
[ dl3 ] compeletly coincides with that of ref .
@xcite which was obtained from the renormalization of the @xmath74 and @xmath75 .
this fact shows another check for the correctness of our calculation .
the finite part is given as follows : _
fn&=&a_1 ^ 2^2(4)^6 ( ^24m^2 ) + & + & ^3 ^ 2(4)^6(^24m^2 ) + & + & a_1^3 ^ 2(4)^6 ^ 2(^24m^2 ) + & + & ^4 ^ 4(4)^6 ( ^24m^2 ) + & + & ^4 ^ 4(4)^6 ^ 2(^24m^2 ) + & + & ^4 ^ 4(4)^6 ^ 3(^24m^2 ) + , [ v3p ] where three dots ( @xmath76 ) denote finite terms which can be absorbed into the counterterms . if we choose the boundary condition @xmath77 , the constant @xmath78 , which is used in obtaining @xmath72 , is determined as c_2=^2(4)^4(112+36)\ { -94++(/24)}. [ c2 ] from the renormalization conditions eq .
( [ rc ] ) we obtain a_3&=&0 , + b_3&=&-2703548 - 754a-25(3 ) + ( -19578 - 92a-6(3))(4 ) + & & -3598 ^ 2(4 ) -278 ^ 3(4 ) + & & + \{-2287251296 - 12536a-12527(3 ) + 1752163 + & & + ( -16769216 - 56a-109(3 ) + 7363)(4 ) -52336 ^ 2(4 ) -98 ^ 3(4 ) } + & & + ^2\{-621053888 - 25162a+1756483 + ( -4775648-a27 + 71083 ) ( 4 ) + & & -319216 ^ 2(4 ) -18 ^ 3(4 ) } + & & + ^3\{-465511664 - 3951944(4 ) -5108 ^ 2(4 ) -1216 ^ 3(4)}.[a3b3 ] putting the constants in eqs .
( [ a12b12 ] ) , ( [ c2 ] ) , and ( [ a3b3 ] ) into eq .
( [ v3p ] ) we finally arrive at the three - loop part of the effective potential : v_eff^(3)(_a)&= & ^3^4 ^ 4(4)^6.[3p ]
in this paper , we have proceeded one loop further in the effective potential calculation of the massless @xmath0 @xmath1 theory , that is , we have calculated the three - loop effective potential of the massless @xmath0 @xmath1 theory in four dimensions for the first time . for an easy check of the cancellation of all the dangerous pole terms among themselves in each loop order
, we have given the @xmath2-expanded loop integrals in full detail .
the @xmath35-order part of eq .
( [ 3p ] ) is our main result .
we see that the order @xmath29- and @xmath34- terms completely agree with previous calculations @xcite in which a different regularization scheme ( cutoff regularization ) was employed . in the cutoff regularization ,
the loop integrations in four dimensions are much more difficult in the three - loop diagrams with the overlap divergence .
we expect the present calculations will serve as a useful reference for a development of the three - loop effective potential of more realistic models containing gauge fields such as the scalar qed . especially in relation to this , the coleman - weinberg mechanism is interesting , which appears in many contexts , from displacive phase transitions in solids to the thermodynamics of the early universe .
this work was supported in part by ministry of education , project number bsri-97 - 2442 and one of the authors ( j. -m .
c. ) was also supported in part by the postdoctoral fellowship of kyung hee university .
in this appendix a , the momenta appearing in the formulas are all ( wick - rotated ) euclidean ones and the abbreviated integration measure is defined as _
k = m^4-n , where @xmath79 is the space - time dimension in the framework of dimensional regularization @xcite and @xmath56 is an arbitrary constant with mass dimension , usually taken as the renormalization scale .
one - loop integrations are quite elementary . for
the two - loop integrations one may refer to ref .
the genuine three - loop integrals which can not be factorized into the lower - loop ones are quite involved . here
we simply quote the results for them .
details of the computation can be found in ref .
@xcite . :
s_0&&_k(1+^2k^2+^2 ) = -(^2+^2)^2(4)^2 + & & ( ^2+^24m^2)^-/2(2 - 2 ) + , + s_1&&_k1k^2+^2 = ^2(4)^2(^24m^2 ) ^-/2(2 - 1 ) , + s_2&&_k1k^2+^2/3 = ^23(4)^2(^2/34m^2 ) ^-/2(2 - 1 ) , + s_3 & & _
k1(k^2+^2)^2=1(4)^2(^24m^2 ) ^-/2(2 ) , + s_4 & & _
k1(k^2+^2/3)^2=1(4)^2(^2/34m^2 ) ^-/2(2 ) . : w_1&&_kp1(p^2+^2)[(p+k)^2+^2 ] = ^4(4)^4(^24m^2 ) ^-^2(2 - 1 ) , + w_2&&_kp1(p^2+^2/3 ) [ ( p+k)^2+^2/3]=^49(4)^4(^2/34m^2 ) ^-^2(2 - 1 ) , + w_3&&_kp1(p^2+^2 ) [ ( p+k)^2+^2/3]=^43(4)^4(^24m^2 ) ^-/2(^2/34m^2)^-/2 ^2(2 - 1 ) , + w_4&&_kp1(k^2+^2 ) ( p^2+^2)[(p+k)^2+^2 ] + & = & ^2(4)^4(^24m^2)^- ^2(1+/2)(1-)(1-/2 ) , + w_5 & & _ kp1(k^2+^2)(p^2+^2/3)[(p+k)^2+^2/3 ] + & = & ^23(4)^4(^2/3 4m^2)^ ^2(1+/2)(1-)(1-/2 ) , + w_6 & & _ kp1(k^2+^2)^2(p^2+^2)[(p+k)^2+^2 ] + & = & 1(4)^4(^24m^2 ) ^- , + w_7 & & _ kp1(k^2+^2/3)^2(p^2+^2/3)[(k+p)^2+^2 ] + & = & 1(4)^4(^2/34m^2 ) ^- , + w_8 & & _ kp1(k^2+^2)^2(p^2+^23)[(k+p)^2+^23 ] + & = & 1(4)^4(^24m^2 ) ^-.[www ] in the above equation , @xmath80 is the usual euler constant , @xmath81 , and numerical values of the constants , @xmath82 , @xmath83 , and @xmath84 in eq .
( [ www ] ) are a = f(1,1)=-1.1719536193 , + b = f(1,3)=-2.3439072387 , + c = f(13,13)=0.1778279325 , where & & f(a , b)_0 ^ 1dx , z = ax+b(1-x)x(1-x ) .
these constants @xmath82 , @xmath83 , and @xmath84 can be analytically integrated @xcite .
the results are expressed in terms of clausen function : a = b2=-32c-34 ^ 2 3= -2cl_2(3 ) , where _
2()=_0^d. : h_1 & &
_ k1(k^2+^2)^2\{_p1p^2+^2}^2 = s_1 ^ 2s_3 , + h_2 & &
_ k1(k^2+^2)^2_p1p^2+^2_q1q^2 + ^2/3=s_1s_2s_3 , + h_3 & & _ k1(k^2+^2/3)^2\{_p1p^2+^2 } ^2=s_1 ^ 2s_4 , + h_4 & & _ k1(k^2+^2)^2\{_p1p^2+^2/3 } ^2 = s_2 ^ 2s_3 , + h_5&&_k1(k^2+^2/3)^2 _ p1p^2+^2_q1q^2+^2/3 = s_1s_2s_4 , + h_6&&_k1(k^2+^2/3)^2 \{_p1p^2+^2/3}^2 = s_2 ^ 2s_4 , + i_1&&_k1k^2+^2 _ p , q1(p^2+^2)^2(q^2+^2)[(p+q)^2+^2 ] = s_1w_6 , + i_2&&_k1k^2+^2/3 _ p , q1(p^2+^2)^2(q^2+^2)[(p+q)^2+^2 ] = s_2w_6 , + i_3&&_k1k^2+^2 _ p , q1(p^2+^2/3)^2(q^2+^2/3)[(p+q)^2+^2 ] = s_1w_7 , + i_4 & & _ k1k^2+^2/3_p , q 1(p^2+^2/3)^2(q^2+^2/3)[(p+q)^2+^2 ] = s_2w_7 , + i_5&&_k1k^2+^2 _ p , q1(p^2+^2)^2(q^2+^2/3)[(p+q)^2+^2/3 ] = s_1w_8 , + i_6&&_k1k^2+^2/3 _ p , q1(p^2+^2)^2(q^2+^2/3)[(p+q)^2+^2/3 ] = s_2w_8 . [ 3 t ] : j_1 & & _ k\{_p1(p^2+^2)[(p+k)^2+^2]}^2 + & & = _ 2 , + j_2 & & _ kpq1(p^2+^2)[(p+k)^2+^2](q^2+^2/3)[(q+k)^2 + ^2/3 ] + & = & _ 2 , + j_3 & & _
k\{_p1(p^2+^2/3 ) [ ( p+k)^2+^2/3]}^2 = _ 2 , + k_1 & & _ k1k^2+^2 \{_p1(p^2+^2)[(p+k)^2+^2]}^2 + & = & _ 1 , + k_2 & & _
kpq1(k^2+^2)(p^2+^2)[(p+k)^2+^2](q^2+^2/3 ) [ ( q+k)^2+^2/3 ] + & = & _ 1 , + k_3 & & _
k1k^2+^2/3 \{_p1(p^2+^2)[(p+k)^2+^2/3]}^2 + & = & _ 1 , + k_4&&_k1k^2+^2 \{_p1(p^2+^2/3)[(p+k)^2+^2/3 ] } ^2 + & = & _ 1 , + l_1 & & _
k1(k^2+^2)^2 \{_p1(p^2+^2)[(p+k)^2+^2]}^2 + & = & _ 0 , + l_2 & & _ kpq1(k^2+^2)^2(p^2+^2)[(p+k)^2+^2](q^2 + ^2/3 ) [ ( q+k)^2+^2/3 ] + & = & _ 0 , + l_3&&_k1(k^2+^2/3)^2 \{_p1(p^2+^2)[(p+k)^2+^2/3]}^2 + & = & _ 0 , + l_4&&_k1(k^2+^2)^2 \{_p1(p^2+^2/3)[(p+k)^2+^2/3 ] } ^2 + & = & _ 0 , + m_1&&_kpq1(k^2+^2)(p^2+^2)(q^2+^2 ) [ ( k - p)^2+^2][(p - q)^2+^2][(q - k)^2+^2 ] + & = & _ 0 , + m_2 & & _
kpq1(k^2+^2/3)(p^2+^2/3)(q^2+^2/3 ) [ ( k - p)^2+^2][(p - q)^2+^2][(q - k)^2+^2 ] + & = & _ 0 , + m_3 & & _
kpq1(k^2+^2)(p^2+^2/3)(q^2+^2/3 ) [ ( k - p)^2+^2/3][(p - q)^2+^2][(q - k)^2+^2/3 ] + & = & _ 0 . [ 3 g ] in the above equation the overall multiplying factors are _ 0=1(4)^6(^24m^2 ) ^ , _
1=^2(4)^6(^24m^2 ) ^ , _ 2=^4(4)^6(^24m^2)^.
diagrams fig . 1 to fig . 3 contain various counterterm integrals in our perturbation theory . in this appendix
b they are shown explicitly : & = & -2(4)^2(m^2 + ( +)^2/21+z)^2(m^2+(+ ) ^2/24m^2(1+z))^-/2(-2 + 2 ) + & = & + & + & ^2 + & + & ^3 , + & = & -(n-1)2(4)^2(m^2 + ( +)^2/61+z)^2(m^2+(+ ) ^2/64m^2(1+z))^-/2(-2 + 2 ) + & = & + & + & ^2 + & + & ^3 , + & = & ^2(+)8(4)^4(1+z)^2 ( m^2+(+)^2/21+z)^2 ( m^2+(+)^2/24m^2(1+z ) ) ^-^2(-1 + 2 ) + & = & ^2 + & + & ^3 , + & = & ^2(n-1)(+)12(4)^4(1+z)^2 ( m^2+(+)^2/21+z ) ( m^2+(+)^2/61+z ) + & & ( m^2+(+)^2/24m^2(1+z ) ) ^-/2 ( m^2+(+)^2/64m^2(1+z ) ) ^-/2 ^2(-1 + 2 ) + & = & ^2 + & + & ^3 , + & = & ^2(n^2 - 1)(+)24(4)^4(1+z)^2 ( m^2+(+)^2/61+z)^2 ( m^2+(+)^2/64m^2(1+z ) ) ^- ^2(-1 + 2 ) + & = & ^2 + & + & ^3 , + & = & -^2(+)^2 ^ 212(4)^4(1+z)^3 ( m^2+(+)^2/21+z ) ( m^2+(+)^2/24m^2(1+z ) ) ^- + & & ( -6 ^ 2 - 3a ) + & = & ^2 + & + & ^3 , + & = & -^2(n-1)(+)^2 ^ 236(4)^4(1+z)^3 ( m^2+(+)^2/61+z ) ( m^2+(+)^2/64m^2(1+z ) ) ^- + & & ( -10 ^ 2 + 63 - 32 ^ 2 3-b ) ] + & = & ^2 + & + & ^3 .
j. van der bij and m. veltman , nucl . phys . *
b231 * , 205 ( 1984 ) ; f. hoogeveen , nucl . phys . * b259 * , 19 ( 1985 ) ; c. ford , i. jack , and d. r. t. jones , nucl . phys . *
b387 * , 373 ( 1992 ) ; a. i. davydychev and j. b. tausk , nucl . phys . *
b397 * , 123 ( 1993 ) ; m. misiak and m. mnz , phys . lett .
* b344 * , 308 ( 1995 ) . | the three - loop effective potential of the massless @xmath0 @xmath1 theory is calculated analytically using techniques of dimensional regularization .
we see a complete agreement between our result and jackiw s result which was obtained only up to two - loop order using a different regularization ( cutoff regularization ) method , but the same renormalization conditions . for an easy check of the mutual cancellation of all the dangerous pole terms in each loop order , we give the @xmath2-expanded loop integrals in full detail .
dratex.sty ~ ps . |
confinement and propagation of surface plasmons in a metal circuitry have received considerable interest for their capability to transport data with a large bandwidth in compact structures and devices . among the different geometries capable of routing the flow of surface plasmon , dielectric - loaded surface plasmon
polariton waveguides ( dlsppws ) @xcite have recently emerged as a potential plasmonic architecture that can be integrated seamlessly with current silicon - on - insulator ( soi ) photonic circuits @xcite and can sustain transfer of information at high date rates @xcite .
a dlsppw is made of a rectangular dielectric rib deposited on a metal film or strip @xcite .
the surface plasmon is confined in the dielectric layer and typical cross - sections required for an optimum confinement of the mode compare well with state - of - the - art soi waveguides operating in the telecom bands . despite dramatically higher losses ,
the advantage of such plasmonic platform is that the optical index of dielectric material used to confined the mode can be externally controlled to realize active dlsppw - based devices @xcite .
understanding device performance such as transmission loss and surface plasmon mode profile greatly contributed to the development of dlsppws .
near - field optical microscopy and far - field leakage radiation microscopy ( lrm ) @xcite are instrumental for visualizing the confinement and propagation details of surface plasmons supported by this geometry @xcite . for thin metal films ,
lrm is an especially useful characterization tool .
it provides a diffraction - limited snapshot of the mode profile in the structure , and , by conoscopic imaging enables to extract the wave - vector distribution .
effective indices of the different modes and interactions developing in a given structure can thus be readily determined @xcite . in this work , we performed an in - depth analysis of leakage radiation images obtained for two different dlsppw - based routing structures : linear couplers and bent waveguides . by simultaneously imaging the conjugated aperture and field planes of the microscope
we unambiguously quantify the degeneracy lift occurring for strongly interacting dlsppws and directly visualize the symmetry of the coupled modes .
furthermore , we measured the wave - vector distribution of 90@xmath0-curved dlsppws and show its evolution with bend radius .
the experimental images are compared to numerical and analytical calculations .
the waveguides considered in this work are depicted schematically in figs .
[ fp_fig1 ] ( a ) to ( c ) .
a poly(methylmethacrylate ) ( pmma ) ridge is defined by electron - beam lithography onto a 65 nm - thick gold film evaporated on a glass substrate .
the thickness of all dlsppws fabricated in this study is fixed at @xmath1=560 nm and the width at @xmath2=600 nm . for such dimensions , the dlsppw structures shown in fig .
[ fp_fig1 ] are single - mode at telecom wavelength @xcite . scanning electron micrograph of the routing elements are shown in figs .
[ fp_fig1 ] ( d ) to ( f ) . of 600 nm and thickness @xmath1=560 nm . ( b ) schematic view of two coupled waveguides separated by a gap @xmath3 .
( c ) schematic view of a curved waveguide with a bend radius @xmath4 .
( d ) , ( e ) and ( f ) are scanning electron micrographs of typical devices corresponding to the configurations ( a ) , ( b ) , and ( c ) , respectively.,width=325 ] = 1510 nm with a long working distance objective .
surface plasmon leakages are collected by an oil immersion objective and are imaged on two cameras placed at the conjugated image ( field ) plane and fourier ( aperture ) plane , respectively.,width=325 ] optical characterization is performed by using a home - made leakage radiation microscope ( lrm ) .
the principle of this method is only recalled here , more detailed can be found in the review by drezet _
_ et al.__@xcite
. the schematic view of our experimental set - up is illustrated in fig .
[ fp_fig2 ] .
an incident tunable laser beam ( fixed at @xmath5=1510 nm hereafter ) is focused by a 100 @xmath6 microscope objective on the extremity of a dlspp waveguide .
the sharp discontinuity defined by the polymer structure acts a local scatterer and produces a spread of wave - vectors , some of which resonant with the surface plasmon modes supported by the geometry . by controlling the incident polarization parallel to the longitudinal axis of the waveguide @xcite ,
the dlsppw mode can thus be readily excited .
leakage radiation microscopy ( lrm ) provides a far - field imaging technique to directly visualize surface plasmon propagation and investigate its fundamental properties @xcite .
this method relies on the collection of radiation losses occurring in the waveguide during propagation .
these losses are emitted in the substrate at an angle phase - matched with the in - plane wave - vector of the spp mode . in our lrm , the plasmon radiation losses are collected by an oil - immersion objective with a numerical aperture ( n.a . ) of 1.49 .
a tube lens focuses the leakages in an image plane ( ip ) conjugated with the object plane where an ingaas infrared camera is recording the two - dimensional intensity distribution .
images recorded at this plane provide direct information about the propagation of the surface plasmon mode developing in the dlsppw . to complete the analysis , we have also used the fourier transforming property of the objective lens to access the wave - vector distribution of the emitted light .
the angular distribution of the rays radiated in the substrate and collected by the objective lens is transformed to a lateral distribution at the objective back focal plane .
quantitative measurement of the complex surface plasmon wave - vector , or equivalently its complex effective index , consists at measuring the radial distance of the rays with respect to the optical axis .
access to momentum space was done by inserting a beam splitter in the optical path and a set of fourier transforming relay lens .
the lenses are forming an image conjugated with back focal plane of the objective . this fourier plane ( fp )
contains the two - dimensional wave - vector distribution of the leakage radiation and momentum - space spectroscopy can thus be readily performed .
( arrow ) .
( c ) exponential fit of the experimental decaying plasmon intensity along the waveguide leading to a @xmath7=42.2 @xmath8 m .
( d ) lorentzian fit of the fp plasmon mode signature centered at @xmath9=1.262 .
the full with at half maximum @xmath10 is inversely proportional to @xmath7.,width=325 ] figures [ fp_fig3](a ) and ( b ) show images of the leakage radiation intensity of a single - mode dlsppw recorded in the conjugated image and fourier planes , respectively .
the plasmon mode is launched at the bottom of the structure and propagates up the waveguide with an exponentially decaying intensity .
the image of fig . [ fp_fig3](b ) is a direct measurement of the wave - vector content of the intensity distribution shown in fig .
[ fp_fig3](a ) . for both fp and ip images ,
a calibration was performed prior to any data extraction . for calibrating ip images , we used the known waveguide length as a standard . with the system magnification one pixel represents @xmath110.6 @xmath8 m . for fp images ,
the @xmath12 of the objective was used to calibrate @xmath13 and @xmath14 axes .
the largest ring in fig . [ fp_fig3](b ) represents the @xmath12=1.49 specified by the manufacturer .
the central disk is the numerical aperture of the @xmath15 excitation objective at @xmath12=0.52 .
one pixel provides a @xmath16 0.012 . the dlsppw mode propagating along this straight waveguide presented in fig . [ fp_fig3 ]
can be defined by two parameters : its effective index @xmath17 and propagation length @xmath7 .
@xmath17 is expressed by the phase constant @xmath18 and reads @xmath19 where @xmath20 .
the propagation length @xmath21 where @xmath22 is the attenuation constant of the plasmon mode .
a complex propagation constant is then evaluated from these two constants : @xmath23 .
@xmath7 can be directly extracted from direct - space image by fitting an exponential decay of the intensity along the waveguide @xmath24 . here
a @xmath7=42.2 @xmath8 m is determined from the fit to the experimental data illustrated in fig .
[ fp_fig3](c ) .
experimentally , the fp image displayed in fig .
[ fp_fig3](b ) contains more information than a direct space image since the real part and imaginary part of the effective index can be directly measured .
the signature of the mode is represented by a single line at a constant @xmath17=@[email protected] .
the intensity measured along @xmath25 at @xmath27=0 is related to the surface plasmon through the following formula @xcite : @xmath28 is the @xmath27-fourier transform of the guided magnetic field at the objective focal point .
the imaginary part of the effective index is also estimated precisely through a lorentzian fit .
the width @xmath10 of the @xmath9 line is a measure of the losses experienced by the plasmon mode and is thus inversely proportional to its propagation length @xmath7 @xcite .
we obtain from the fp image a @xmath7=43.3 @xmath8 m in fairly good agreement with @xmath7=42.2 @xmath8 m measured from direct - space analysis .
equals to 670 nm , 600 nm , 440 nm , 350 nm , 240 nm and 90 nm respectively .
( h ) to ( m ) are the corresponding wave - vector distributions unambiguously demonstrating the degeneracy lift of the coupled waveguides for small values of @xmath3 .
the images were cropped to display only the lower part of the fourier plane.,width=325 ] we now demonstrate the added - value of performing momentum - space leakage radiation spectroscopy of a classical integrated plasmonic routing device : a linear dlsppw coupler @xcite . the linear coupler geometry illustrated in fig . [
fp_fig4](a ) consists of two parallel and identical dlspp waveguides separated by an edge - to - edge distance @xmath3 varying from 670 nm to 90 nm .
this elementary configuration is well - known from coupled - mode theory @xcite .
when the gap distance @xmath3 is reduced the degenerate modes propagating in the uncoupled waveguides are split into symmetric and antisymmetric modes .
new propagation constants @xmath29 and @xmath30 , respectively are thus characterizing the symmetric mode and the antisymmetric mode , respectively , and they critically depend on @xmath3 . a beating of these two modes can be observed in leakage radiation microscopy @xcite where a mode propagating in one waveguide can be totally transferred to the second after a coupling distance @xmath31 @xcite .
figures [ fp_fig4](b)-(g ) qualitatively show the evolution of the beating pattern and the coupling distance @xmath31 for decreasing separation distances @xmath3 .
more interesting are the corresponding wave - vector distributions depicted in the series of fp images in figs .
[ fp_fig4](h ) to ( m ) . when the dlsppws are weakly coupled ( @xmath3=670 nm ) , the fourier content of fig .
[ fp_fig4](h ) strongly resembles that of a single dlspp mode already shown in fig .
[ fp_fig3](b ) .
when @xmath3 is reduced , a clear splitting of the modes is observed indicative of a strong interaction between the waveguides .
the parity of the modes can be readily determined from _
e.g. _ fig .
[ fp_fig4](m ) .
the asymmetric mode has an odd parity with two maxima centered on each side of @xmath25 axis .
figure [ fp_fig5 ] illustrates the benefit of performing momentum - space spectroscopy described here .
figures [ fp_fig5](a ) and ( b ) are leakage radiation images recorded at the conjugated fourier plane and image plane for a linear coupler with @xmath3=440 nm , respectively .
the effective indices of the symmetric @xmath32 and antisymmetric @xmath33 modes are evaluated by lorentzian fits of crosscuts of the momentum distribution along the @xmath25 axis marked by the circles in fig .
[ fp_fig5](a ) .
@xmath33 was evaluated at two different wave - vector positions with respect to the @xmath25 axis , labeled as @xmath33 and @xmath34 .
the fits to the data are represented in figs .
[ fp_fig5](c ) and ( d ) for the antisymmetric mode leading to @[email protected]
. the effective index of the symmetric mode is measured at @[email protected] ( fig .
[ fp_fig5](e ) ) .
the coupling length @xmath31 can now be evaluated using the following relation @xcite @xmath37 where @xmath38 and @xmath39 .
then @xmath40 this procedure was repeated for the different dlsppw separation distances @xmath3 .
the extracted values of the split modes and coupling length @xmath31 are reported in table [ gvar ] .
@xmath41 was estimated taking into account an average value of @xmath33 and @xmath34 . for comparison purposes ,
@xmath31 was also determined by analyzing the beating pattern recorded in the image plane ( fig .
[ fp_fig5](b ) ) . a longitudinal cross - section of the leakage intensity taken along each dlsppws is shown in fig .
[ fp_fig5](f ) for @xmath3=440 nm .
the transfer of the mode between the two waveguides is obtained after @xmath4216 @xmath8 m . within our pixel resolution ,
this value compares well with the @xmath31 determined by the analysis of the degeneracy lift of the effective indices by eq .
the advantage of a momentum - space spectroscopy is that , unlike near - field measurement @xcite , the effective indices can be directly measured and mode symmetry visualized @xcite . @xmath31
inferred from direct plane analysis only provides the difference between the two propagation constants @xmath43 without discriminating the symmetric mode from the antisymmetric one .
= 440 nm .
( c ) and ( d ) are lorentzian fits of the asymmetric mode at the location marked by the circle in ( a ) .
( e ) is a lorentzian fit of the symmetric mode .
( f ) longitudinal intensity cross sections taken along the two coupled waveguides in ( b ) showing the energy transfer from one dlsppw to the other defining the coupling length @xmath31.,width=325 ] .[tab1 ] measured values of the effectives indices of the asymmetric and symmetric modes and estimated coupling length @xmath31 for linear dlsppw couplers with different edge - to - edge separation @xmath3 .
@xmath31 values were estimated independently by momentum - space spectroscopy and direct - plane analysis . [ cols="^,^,^,^,^,^",options="header " , ] to confirm these experimental results , we numerically simulated linear dlsppw couplers with the commercial finite - element mode solver comsol .
the optical index of the pmma waveguide is @xmath44 and that of the gold layer is @xmath45 at @xmath5=1510 nm @xcite .
the transversal electric field distribution of the asymmetric and symmetric mode are shown figs .
[ fp_fig6 ] ( a ) and ( b ) for @xmath3=400 nm .
figure [ fp_fig6](c ) provides a comparison of the evolution of the degeneracy lift with separation distance @xmath3 .
the measured data quantitatively reproduce the response of the simulated device .
= 400 nm .
( c ) comparison between the experimental splitting and the calculated values for various coupling distances @xmath3.,width=325 ]
in this section , we investigate the momentum distribution of another well - known basic routing element : 90@xmath46 curved waveguides . this dlsppw structure has been extensively studied by various groups and the effect of bend radius on the overall losses
is well understood @xcite . by performing momentum - space spectroscopy of the supported mode
, we visualize and analyze the wave - vector content of the bend section for this routing device .
we demonstrate the limitations of momentum - space spectroscopy to extract modal properties of this elementary building block .
the curved dlsppw is composed of two @xmath47=30 @xmath8 m long straight waveguides with @xmath2=600 nm .
these two input and output waveguides are linked by a circular @xmath48 bend section of radius @xmath4 .
bent dlsppws with @xmath4 ranging from 5 @xmath8 m to 19 @xmath8 m were fabricated .
the leakage intensity distribution image of a @xmath4=19 @xmath8 m curved waveguide is shown fig .
[ fp_fig7](a ) .
the corresponding wave - vector distribution is depicted in fig [ fp_fig7](b ) .
the two straight lines at @xmath27 and @xmath49 constant are related to the mode propagating along the @xmath50 and @xmath51-oriented waveguides , respectively .
the line at @xmath27=constant is the input waveguide .
this fourier plane shows additional signatures such as the illumination wave - vector span ( central disk ) and planar plasmon modes supported by the au / air or au / pmma taper interfaces .
of particular interest here is the fourier signature of the bent section of the waveguide visible as an arc of circle linking the two @xmath27 and @xmath49 lines .
= 19 @xmath8 m and its corresponding wave - vector distribution , respectively .
the fourier content of the curved section is appearing as an arc linking the two @xmath27 and @xmath49 lines .
( c ) and ( d ) computed intensity of the mode propagating in the routing element and its corresponding wave - vector distribution , respectively.,width=325 ] instead of full numerical simulations , we propose in the following a simple analytical model that explains the main features of the measured fourier plane images . to this aim , we approximated the mode that propagates in the bend structure by a gaussian profile with a finite propagation length .
the characteristics of the gaussian profile was defined by the experimental data obtained from a straight waveguide ; namely @xmath17=1.262 and @xmath7=42.2 @xmath8 m .
the mode width was fixed at @xmath52=500 nm according to the analysis of holmgaard and bozhevolnyi @xcite .
the magnetic field is written as follow : * input straight guide ( @xmath53 , @xmath54 ) @xmath55\exp [ i \beta x]\ , , \\ & & = h_o\exp\left[\frac{-(y - r_c)^2}{w_0 ^ 2}\right]\exp [ i n_{eff}k_o x ] \exp [ -x/2l_{spp } ] \ , , \nonumber\end{aligned}\ ] ] * circular portion ( @xmath56 , @xmath54 ) @xmath57\\ & & \exp\left[\frac{-(r - r_c)^2}{w_0 ^ 2}\right]\exp[i k_b r_c \theta ] \ , , \nonumber\end{aligned}\ ] ] where @xmath58 , @xmath59 and @xmath60 is the complex wave - vector of the mode in the curved section . * output straight guide ( @xmath56 , @xmath61 ) @xmath62\exp[-l/2l_{spp}]\\ & & \exp[-r_c\pi/4l_{spp}]\exp [ i n_{eff}k_o y ] \exp[-y/2l_{spp}]\ , . \nonumber\end{aligned}\ ] ] the momentum representation @xmath63 was obtained by a fourier transform @xmath64 \mathrm{d } x \mathrm{d } y. \nonumber\end{aligned}\ ] ] where the integration window is truncated to limit the calculation on the mode extension . outside this area , the mode profile vanishes so that its contribution to the fourier transform is negligibly small . to simulate the curved dlsppw
, we maintained the complex propagation constant in the curved section equal to that of the straight waveguides ( @xmath65 ) .
this assumption remains valid for @xmath66 where @xmath67 is the limiting radius where bend losses can be neglected .
@xmath68 have been recently numerically estimated for dlsppw @xcite confirming an analytical expression historically used for standard optical waveguides @xcite @xmath69 here @xmath70 corresponds to the length over which the field outside of the waveguide decays by @xmath71 . with @xmath72 ,
@xmath73 13 @xmath8 m .
when the radius is below @xmath67 , bending losses are induced by a radial displacement of the mode profile with respect to the waveguide axis @xcite .
this displacement pushes the field outside the waveguide leading to a lower phase velocity and a modification of the effective index @xcite . considering curvature loss
as an additional exponential decay is a good approximation to quantify the total loss induced by the bend and estimate the transmission of the 90@xmath0 waveguide . however ,
this approximation is not representative of the real shape of the field along the bend and consequently can not be used to model fourier images of the kind displayed in fig .
[ fp_fig7](b ) . for @xmath7413
@xmath8 m , we show that the propagation and momentum - space representations simulated with these basic assumptions ( figs .
[ fp_fig7](c ) and ( d ) ) reproduces the experimental observations of figs .
[ fp_fig7](a ) and ( b ) .
we compare the profile of the calculated fourier transform ( fig .
[ fp_fig7](d ) ) to the experimental wave - vector distribution ( fig .
[ fp_fig7](b ) ) by extracting momentum profiles at different radial coordinates indicated in fig .
[ fp_fig8](a ) for @xmath4=19 @xmath8 m .
figures [ fp_fig8](b ) , ( c ) and ( d ) show the calculated and experimental wave - vector distributions taken along the @xmath27 and @xmath49 axis and at -@xmath75 , respectively .
the position and width of the calculated wave - vector profiles are in very good agreement with the measured data validating thus our preliminary assumptions .
some extra signatures are visible on the experimental cuts : a peak at 1.49 in fig .
[ fp_fig8](b ) corresponds to the numerical aperture of the collection objective , and the contribution of the surface plasmon excited in the nearby gold film is visible at 1.08 in fig .
[ fp_fig8](d ) .
noticeable also is the presence of gibbs oscillations revealed by this momentum - space spectroscopy ( arrows figs .
[ fp_fig8 ] ( b ) and ( c ) ) .
the origin of these oscillations in reciprocal space is discussed below . in order to provide analytical expressions and propose a simple physical understanding of the measured fourier images , we simplified further the model .
we consider each part of the waveguide ( straight and bend ) independently from each others .
the straight parts are approximatively 30 @xmath8 m long , a length smaller than the longitudinal decay of the supported mode ( @xmath7=42 @xmath8 m ) .
if we consider only the field in the input straight waveguide along the @xmath51 coordinate , the fourier transform of the magnetic field written in eq .
[ eqstraight ] reads @xmath76\ , \mathrm{d } x \,,\ ] ] where we neglected the @xmath50-gaussian profile for the sake of clarity but this could be easily considered .
then the intensity in the fourier plane along @xmath27 direction writes @xmath77 if @xmath78 then @xmath79 $ ] and follows a lorentzian profile , as expected .
if the straight part length @xmath47 is smaller than @xmath7 , the resulting fourier transform is not simply defined by a lorentzian function .
because of the finite integration limit , extra oscillations are becoming visible ( gibbs oscillations ) as illustrated in figs .
[ fp_fig8](b ) and ( c ) . for long waveguide ( as in fig .
[ fp_fig3](a ) where @xmath80 ) , no oscillation of the wave - vector distribution is observed .
figure [ fp_fig9 ] shows the impact of the integration boundaries @xmath47 on the momentum space representation of a single straight waveguide for a fixed propagation length of 40 @xmath8 m . there is a perfect agreement between the numerically calculated fourier transform and its analytical solution given by eq .
[ eq9 ] . for integration boundaries greater than the propagation length
the two calculations match with the lorentz function generally used . and
crosses : analytical solutions ( as ) calculated from eq . [ eq9 ] with the same integration boundaries .
red solid line : calculated lorentzian profile of the wave - vector distribution of a plasmon mode with a propagation length of 40 @xmath8 m .
when @xmath82 gibbs oscillations and widening of the wave - vector content occur.,width=325 ] since the momentum space representation of a straight waveguide can be expressed for any integration boundaries , we now look for the expression of a curved waveguide alone . on fig .
[ fp_fig8 ] ( d ) , the @xmath83 profile shows the experimental momentum space of a @xmath48 curved waveguide linked by an input and an output waveguide .
we approximate in the following a solution considering a fully symmetric configuration consisting of a lossless circular waveguide .
losses are omitted to maintain the radial symmetry valid between [ 0;2@xmath84 .
the fourier transform in a @xmath85 bend can be expressed in polar coordinates as @xmath86 @xmath87 is given by eq .
since this function is periodic with respect to @xmath88 between [ 0;2@xmath84 , we can expand @xmath87 in a fourier series : @xmath89 where the @xmath90 harmonic is @xmath91 } { \rm sinc}(k_b r_c \pi - n\pi ) .
\nonumber\end{aligned}\ ] ] the fourier transform of the field ( eq . [ eq10 ] ) can now be written as follow @xmath92 @xmath93 with @xmath94 and @xmath95}j _ { n } ( 2\pi k _
{ r } r ) } dr.\nonumber\ ] ] @xmath96 is the bessel function of the first kind of @xmath90 order .
further simplification can be achieved by extracting the main @xmath90 harmonic @xmath97 of the series .
it is deduced from eq .
[ eq14 ] and is such that @xmath98 therefore the fourier series is dominated by the harmonic integer that is the closest to the product @xmath99 .
= 19 @xmath8 m , 13 @xmath8 m , 9 @xmath8 m , and 7 @xmath8 m , respectively .
( e ) to ( h ) experimental profiles ( shaded areas ) of the wave - vector distribution taken along the dashed line in ( a ) to ( d ) .
the solid blue lines are the calculated profiles using eq .
[ eq : ft ] already reported in fig .
[ fp_fig8 ] .
the red dashed lines show the analytical solutions obtained with the @xmath90 order rendering the best agreement with the data .
@xmath100=100 , 68 , 43 , and 37 , respectively.,width=325 ] figure [ fp_fig10 ] illustrates a comparison between the experimental data extracted from momentum - space spectroscopy , the fourier calculations derived from eq .
[ eq : ft ] already used in figure [ fp_fig8 ] , and the analytical approximation discussed above .
the experimental fourier planes of figs .
[ fp_fig10 ] ( a ) to ( d ) only show the region of interest for four different radii .
the corresponding wave - vector profiles extracted and calculated at @xmath101/4 ( dashed lines in figure [ fp_fig7](d ) ) are plotted in the graphs of figs .
[ fp_fig10 ] ( e ) to ( h ) . for radius @xmath66 ,
the fourier calculations and the analytical solutions are in good agreement with the experimental data . for large radii
the signature of the bend in momentum space is defined only by the phase difference @xmath102 .
this corresponds to resonance condition of a ring resonator .
experimentally , the periodic condition on @xmath88 is not respected since the structure considered is formed by an arc of circle . nonetheless , using the @xmath103 order of the bessel function in eq .
[ eq13 ] for @xmath4=19 @xmath8 m and the @xmath104 order for @xmath4=13 @xmath8 m the experimental data and the calculated fourier transform can be well reproduced ( figs .
[ fp_fig10 ] ( e ) and ( f ) ) .
for radius @xmath105 , the calculated profiles deduced from eq .
[ eq : ft ] and from the @xmath106 and @xmath107 orders for @xmath4=9 @xmath8 m and @xmath4=7 @xmath8 m , respectively are deviating from the experimental cross - cuts ( figs .
[ fp_fig10 ] ( g ) and ( h ) ) .
this disagreement is expected since none of the two models ( fourier transform and analytical approximation ) are including bending loss .
by using dual - plane leakage radiation microscopy we have fully quantified the key parameters characterizing two important dielectric - loaded surface plasmon polariton routing devices : linear couplers and 90@xmath46 curved waveguides .
we unambiguously demonstrated the added - value of performing momentum - space spectroscopy .
the degeneracy lift for strongly coupled waveguides and the symmetry of the split modes can be directly visualized and quantified .
the wave - vector distribution associated to the curved section of the waveguide was also readily observed .
we developed a numerical and an analytical analyzis to understand the experimental momentum distribution .
we found that for large radii ( vanishing bending loss ) , we can link the plasmon signature in fourier space with the geometrical and modal properties of the bend structure .
the radial dependence of the wave - vector distribution is governed by the phase difference @xmath108 . for smaller radii of curvature
the bend loss need to be accounted for by developing an approach including realistic field shape .
this work was funded by the european fp7 research program platon , contract number 249135 and the european research council grant agreement number 306772 , the regional council of burgundy under the program pari smt3 and the labex action .
s. lachze is acknowledged for participating at the early stage of this work , with the support of burgundy region and cea leti carnot funding .
34ifxundefined [ 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 @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) http://dx.doi.org/10.1021/nl901314u [ * * , ( ) ] link:\doibase 10.1364/oe.18.001207 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1364/ol.36.002731 [ * * , ( ) ] link:\doibase 10.1364/oe.19.025222 [ * * , ( ) ] link:\doibase 10.1063/1.3609081 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1063/1.3525160 [ * * , ( ) ] link:\doibase 10.1364/oe.19.005303 [ * * , ( ) ] link:\doibase 10.1364/oe.20.008658 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ surfaces and on gratings _ , , vol .
( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) , ed . ,
@noop _ _ ( , ) @noop * * , ( ) link:\doibase 10.1364/oe.18.016112 [ * * , ( ) ] http://stacks.iop.org/0256-307x/27/i=2/a=027303 [ * * , ( ) ] link:\doibase 10.1364/oe.20.023898 [ * * , ( ) ] @noop _ technology _ ( , ) | we perform advanced radiation leakage microscopy of routing dielectric - loaded plasmonic waveguiding structures . by direct plane imaging and momentum - space spectroscopy
, we analyze the energy transfer between coupled waveguides as a function of gap distance and reveal the momentum distribution of curved geometries .
specifically , we observed a clear degeneracy lift of the effective indices for strongly interacting waveguides in agreement with coupled - mode theory .
we use momentum - space representations to discuss the effect of curvature on dielectric - loaded waveguides .
the experimental images are successfully reproduced by a numerical and an analytical model of the mode propagating in a curved plasmonic waveguide . |
two - point correlators of heavy quarks are very useful objects for extracting parameters of qcd such as , e.g. , the quark masses and the strong coupling constant @xmath14 , both from experiment@xcite and also from the lattice@xcite . however , even our perturbative knowledge of these functions is quite limited due to the complexity of the required diagrammatic calculation .
for instance , in the case of the vector vacuum polarization function , only at @xmath15 and @xmath16 is the analytic result fully known @xcite . at @xmath17 ,
state - of - the - art techniques produce partial answers for this function only in the form of local expansions around three particular values of the external momentum , to wit @xmath0 , @xmath3 and @xmath18 where @xmath2 is the heavy - quark mass .
these correspond to the so - called low - energy expansion , the operator product expansion and the threshold expansion of the correlator .
a phenomenal effort has been devoted to the calculation of an ever - increasing number of coefficients in all of these expansions .
for instance , for the @xmath17 contribution to the vacuum polarization , no less than 30 terms are known from the low - energy expansion @xcite ; although in most of all other situations this number is reduced to just a few @xcite . ideally , of course
, one would like to be able to know the full function ( including its imaginary part ) and , consequently , a strategy has to be devised in order to reconstruct the function from the three expansions , even if this is only approximate . up to now ,
the method that has been employed in this reconstruction is that of pad approximants @xcite .
pad approximants are ratios of two polynomials whose coefficients are tuned so that they reproduce the taylor expansion in @xmath10 around the origin of a given correlator , to the highest possible degree .
however , even though the vacuum polarization is an analytic function and , therefore , admits a taylor series around @xmath0 , this is not so either at threshold or for @xmath19 due to the presence of non - analytic dependence in the form of square - roots and logarithms of the momentum .
consequently , all by themselves , pad approximants can not approximate the correlator in the full complex plane , unless this nonanalytic dependence is somehow removed . in order to do this ,
a battery of judiciously chosen functions is used for subtracting the nonanalytic dependence on @xmath10 from the original correlator .
pad approximants are then constructed for the subtracted correlator , rather than for the correlator itself . as pointed out in , e.g. , ref .
@xcite this procedure has a certain degree of arbitrariness .
this ambiguity in the procedure can potentially be the source of a systematic uncertainty , and it is important that this uncertainty be quantified .
normally , this is attempted by varying among several of the choices made in the aforementioned pad construction .
the results obtained with the pad method are certainly interesting but , as higher and higher precision is always in demand , we think it is also interesting that one could devise alternative methods which could confirm ( or if need be , modify ) these results . therefore
, we think it would be desirable to have an alternative procedure which could reconstruct the full function from the three expansions in an unambiguous way , and with which we could have a better handle on the possible systematic uncertainties involved . in this paper
we will present what we believe to be one such method . furthermore , unlike pads which are essentially numerical , our reconstruction is totally analytic .
the main observation is that , in the conformal variable @xmath7 in which the @xmath10 cut complex plane is mapped onto a unit disk , the coefficients of the threshold expansion and the ope determine the asymptotic behavior of the coefficients of the low - energy expansion .
this property is easiest to recognize in the language of mellin transforms , thanks to a mathematical result which is known as the converse mapping theorem @xcite .
an intuitive way to understand this is the following .
a finite set of taylor coefficients knows nothing about the region of analyticity of the function .
so , for example , even if you know a thousand taylor coefficients , strictly speaking , you still do not know the value for the radius of convergence .
the reason is because the radius of convergence is a limit procedure and things can change completely from the coefficient a thousand and 1 onwards .
since the region of analyticity knows about the asymptotic tail of the taylor coefficients , it is not strange that by giving precise information on the properties of the function at the boundary of this region of analyticity ( i.e. the threshold expansion and the ope ) one may gather information on this asymptotic tail of the taylor coefficients .
how to make this statement useful beyond just words , is what we have tried to do in this work . as it turns out , the asymptotic behavior of the low - energy coefficients sets in so quickly that already for the second or the third coefficient it is a good approximation .
this results in an efficient way to compute them .
once all the taylor coefficients are known ( even if only approximately ) one can resum them and reconstruct the full function . as we will see
, this resummation expresses the result for the full function as a unique combination of polylogarithms ( and derivatives thereof ) plus a known polynomial .
although we have set up the problem within the context of a two - point correlator of a heavy quark , we think that the method we have developed can be of more general use , and could be applied also in other contexts where several expansions to a function with a given analytic structure are known .
mellin transforms have also proven very useful in other contexts in perturbative calculations @xcite . in the next section
we will set up the necessary definitions and describe in detail our method and the relevant mathematical results it is based on . in section 3 , we will test the method with the case of the @xmath17 vacuum polarization function which , although not fully known analytically , is sufficiently known to allow interesting nontrivial checks . in particular , since 30 low - energy coefficients have been calculated up to now @xcite , we will be able to compare our prediction for them with their exact value . in section 4
, we will apply the method to the @xmath20 vacuum polarization function , for which only a few terms in the low - energy , threshold expansion and ope have been calculated @xcite .
we will reconstruct the full function , make again predictions for the coefficients of its low - energy expansion , and compare them with those already present in the literature .
the final section will be devoted to some conclusions and outlook .
finally , in the appendix , we compare our approximation to the exact result in the case of a known function , choosing the result at @xmath15 as an illustrative example .
let @xmath21 be a function of the complex variable @xmath22 with the following properties : * it is analytic in the disc @xmath23 where , consequently , it admits a convergent expansion when @xmath24 as + @xmath25 + which , from now on , we will call the `` taylor '' expansion .
* it has a cut along @xmath26 . in the limit @xmath27
, it can be expanded in the form + @xmath28 + where @xmath29 are integers but @xmath30 may be integers or half - integers . this expansion will be called the `` threshold '' expansion . * finally , it admits an expansion for @xmath31 such as @xmath32 where @xmath30 and @xmath29 are both integers .
we will refer to this expansion as the `` ope '' .
the problem we are trying to solve is how to reconstruct the function @xmath33 which matches onto the above three expansions in the corresponding regions , in an analytic and systematic way . that this is possible is already guaranteed by the taylor expansion ( which is convergent ) and the theorem on analytic continuation
. however , an exact reconstruction requires complete knowledge of the infinite set of coefficients @xmath34 . realistically speaking ,
this type of information is certainly not available .
when only a finite number of @xmath34 s is known , one clearly has no other option but to resort to the ope and threshold expansions to try to reconstruct the original function @xmath33 .
it seems reasonable to imagine that the more terms one knows , the better one will be able to determine this function .
however , to the best of our knowledge , a concrete method to analytically reconstruct @xmath33 is not available in the literature .
therefore , we would like to explain how one can do this reconstruction in an analytic and systematic way .
this we will do with the help of the mellin transform , using the vector two - point correlator for a heavy quark as an example , and illustrating how the method works in practice .
let us recall the definition of the vacuum polarization function @xmath35 through the correlator of two electromagnetic currents @xmath36 , where @xmath37 is a heavy quark , @xmath38 and where @xmath39 is the external four - momentum . in qcd perturbation theory ,
@xmath35 may be decomposed to @xmath40 as @xmath41 for definiteness , @xmath14 denotes the strong coupling constant in the @xmath42 scheme at the scale @xmath43 , but the precise definition is not important for the discussion which follows .
equation ( [ pi ] ) will be understood in the on - shell normalization scheme where a subtraction at zero momentum has been made in such a way as to guarantee that @xmath44 . as it is well known , the vacuum polarization in eq .
( [ pi ] ) satisfies a once subtracted dispersion relation , i.e. @xmath45 when massless cuts are disregarded ( as we will assume henceforth ) , the spectral function in ( [ disprel ] ) starts at @xmath46 , where @xmath2 is the heavy quark pole mass . in this case
, the function @xmath47 may be expanded for @xmath48 in powers of @xmath49 and it admits a taylor expansion , exactly as in eq .
( [ taylorpi ] ) .
the other two expansions , eqs .
( [ thresholdpi],[opepi ] ) coincide with the threshold expansion ( for @xmath50 ) and the ope ( when @xmath19 ) . in order to bring the three expansions ( [ taylorpi]-[opepi ] ) to a common ground , it is very convenient to make the conformal transformation @xmath51 which maps the cut @xmath22 plane into a unit disc in the @xmath7 plane , as we can see on figure [ omegaplan ] .
the cut @xmath52 is transformed into the circle @xmath53 and the points @xmath54 into @xmath55 , @xmath56 into @xmath57 , and the limit @xmath58 into @xmath59 , and @xmath60 into @xmath61 . and @xmath7 , eq . ( [ conformal ] ) .
, width=576 ] in terms of the new variable @xmath7 , the vacuum polarization function @xmath62 also obeys the first and second expansions ( [ taylorpi ] , [ thresholdpi ] ) , while the ope becomes formally like the threshold expansion , except for a flip of sign in @xmath7 . to wit : * the taylor expansion becomes + @xmath63 * whereas the ope ( @xmath64 ) and threshold ( @xmath65 ) expansion become + @xmath66 the two sets of coefficients , @xmath67 and @xmath68 are , in principle , unrelated as they correspond to expansions of the same function at two different points .
however , one can find the connection between the two sets of coefficients @xmath34 in eq .
( [ taylorpi ] ) and @xmath6 in eq .
( [ taylorpiw ] ) .
they read @xmath69 as we will now see , the coefficients @xmath70 in ( [ thresholdpiw ] ) determine the behavior of @xmath6 for large values of @xmath8 .
in order to see this , it is very convenient to redefine the conformal variable @xmath7 as @xmath71 the limit @xmath72 with the `` @xmath73 '' sign corresponds to the threshold expansion , whereas the `` @xmath74 '' sign corresponds to the ope .
then , one can make use of the integral representation @xmath75 where the integral on the righthand side term runs over a straight line in the @xmath76 complex plane , with @xmath77 any constant real number , @xmath78 .
the second identity allows one to express the vacuum polarization function ( [ taylorpiw ] ) as an inverse mellin transform in the @xmath79 variable as @xmath80 in this expression the integration path is a vertical line in the @xmath76 complex plane located at @xmath81 , within the region of full analyticity of the integrand .
this region is usually called `` the fundamental strip '' .
the usefulness of the expression ( [ mellinpit ] ) for our purposes is due to a relatively recent mathematical theorem , known as the _ converse mapping theorem _ @xcite , which assures that the @xmath72 behavior of the function @xmath82 is completely determined by the singularities in the @xmath76 complex plane of the function appearing in the integrand of ( [ mellinpit ] ) , i.e. @xmath83 furthermore , these singularities are just limited to isolated poles , i.e. the above combination @xmath84 is a meromorphic function . in general , the converse mapping theorem establishes the following dictionary @xcite between a given function @xmath85 and its mellin transform @xmath86(s ) $ ] : @xmath87(s ) & = \int_0^\infty\ ! dt \ ; t^{s-1 } \ ; f(t ) & \rightleftharpoons & & f(t ) & = \int \limits_{c- i \infty}^{c+ i \infty } \ !
\frac{ds}{2i\pi } \ ; t^{-s } \ ; \mathcal{m}[f](s ) \label{dict1}\\ \mathcal{m}[f](s ) & \asymp \sum_{p , k } \frac{r_{p , k}}{(s+p)^k } &
\rightleftharpoons & & f(t ) & \underset{t\rightarrow0}{\sim } \sum_{p , k } \frac{(-1)^{k-1}}{(k-1)!}\ ; r_{p , k}\ ; t^p \log^{k-1 } t \;. \label{dict2}\end{aligned}\ ] ] in the expression above the symbol @xmath88 means `` singular expansion '' @xcite , i.e. the sum of all the negative powers of the laurent expansions of the function around every pole .
in particular , we would like to call the reader s attention to the result shown in eq .
( [ dict2 ] ) , which says that the expansion of the function at @xmath72 is given by the pole @xmath30 ( together with its multiplicity @xmath29 ) , and the residue @xmath89 , of its mellin transform .
this property will allow us , for instance , to compute the behavior of the coefficients @xmath6 in the expansion ( [ taylorpiw ] ) for @xmath8 large .
one must first realize that the position and residue of a pole of the @xmath90 function is entirely given by the asymptotic behavior as @xmath91 of the @xmath6 coefficients .
this is very important since it means that the function @xmath92 admits an expansion in powers of @xmath9 for @xmath8 large .
for instance , if @xmath93 one finds that @xmath94 where @xmath95 is the @xmath29-th derivative of the riemann @xmath96 function , and @xmath97 is the @xmath29-th derivative of the dirichlet @xmath98 function , @xmath99 now , @xmath100 exhibiting the pole at @xmath101 .
so the only pole of @xmath102 in eq .
( [ asympt ] ) is located at @xmath103 .
one immediately sees that the location of this pole is correlated with the power of @xmath8 in the large-@xmath8 behavior of the @xmath6 coefficient ( [ example ] ) , while the multiplicity of the pole is given by the power of the @xmath104 . on the contrary
, the dirichlet @xmath98 function has no pole .
of course the results in @xmath102 and @xmath105 get interchanged if the @xmath6 coefficient has the same large-@xmath8 behavior but with an alternating sign .
looking at the vacuum polarization function @xmath106 in eq .
( [ mellinpit ] ) , one sees that one goes from the threshold expansion to the ope by the addition of the alternating sign in the definition of the function @xmath90 .
this means that both expansions will be simultaneously encoded in a @xmath6 coefficient admitting an expansion for large @xmath8 of the form @xmath107\frac{\log^k n } { n^p}\ ; . \label{defomega}\ ] ] in fact , the coefficients @xmath108 encode the information from the non - analytic terms of the threshold expansion , and the @xmath109 the information from the non - analytic terms in the ope , as we will see explicitly in the next sections . to illustrate the procedure , we will first treat the case of the @xmath110 contribution to the vacuum polarization , eq .
( [ pi ] ) . since no less than 30 terms are known in the taylor expansion of this function ( [ taylorpi ] ) , this information will be very helpful to check our results and our understanding of the systematic errors involved .
furthermore , we will approximately reconstruct the function in the complex plane , including its imaginary part , and will be able to compute the value of a constant in the threshold expansion of this function which , up to now , has not been possible to obtain by ordinary diagrammatic methods . after that , we will move to the more interesting ( and also more difficult ) case of @xmath111 , where we will repeat the same analysis , except that now only 3 coefficients from its taylor expansion ( [ taylorpi ] ) together with just a few from the ope and threshold expansion , are currently known .
let us begin by considering the function @xmath112 .
the threshold expansion ( [ thresholdpi ] ) has been calculated in @xcite to be @xmath113 in terms of certain known coefficients which we denote here as @xmath114 .
for instance , @xmath115 is the constant called @xmath116 in reference @xcite . in terms of the conformal variable @xmath7 ( [ conformal ] ) , this expansion reads ( [ thresholdpiw ] ) @xmath117 obviously the two expansions ( [ pi2threshold ] ) and ( [ pi2omegaplus ] ) are related , with @xmath118 as we have discussed ( [ deft ] ) , it is convenient to re - express the expansion ( [ pi2omegaplus ] ) in the @xmath79 variable , ( recall that @xmath119 ) , as @xmath120 this means , through the _ converse mapping theorem _ ( [ dict1],[dict2 ] ) , that the mellin transform is given by @xmath121(s ) = \gamma(s)\ g_+(s){\nonumber}\\ & \asymp \frac{\omega^{(+)}(-1,0)}{s-1}\ + \ \frac{2\ \omega^{(+)}(0,2)}{s^3}\ -\ \frac{\omega^{(+)}(0,1)}{s^2}\ + \ \frac { \omega^{(+)}(0,0 ) + \frac{1}{2}\ \omega^{(+)}(-1,0 ) } { s } { \nonumber}\\ & \;\;\ ; + \ \frac { \omega^{(+)}(0,2 ) - \omega^{(+)}(1,1 ) } { ( s+1)^2 } + \ \frac { \frac{1}{12}\ \omega^{(+)}(-1,0)-\frac{1}{2}\ \omega^{(+)}(0,1)+\omega^{(+)}(1,0 ) } { s+1}+ \ldots\ ; , \label{mellinthreshold}\end{aligned}\ ] ] where the last expression corresponds to its singular expansion
. as one can see , the residues in this singular expansion are given by the coefficients of the threshold expansion ( [ pi2omegaplus ] ) in a very specific way .
turning now to the ope ( [ opepi ] ) , we have @xcite @xmath122 where the constants @xmath123 and @xmath124 are called @xmath125 and @xmath126 in reference @xcite .
in fact , many more terms of the ope are known @xcite , but we stop at @xmath127 because it is enough for our illustrative purposes . in terms of the conformal variable @xmath7 ( [ conformal ] ) , one has @xmath128 where , again , the two expansions are related through @xmath129 and similar expressions for @xmath130 and @xmath131 which , because of their length , we do not write out explicitly . finally , in terms of the @xmath79 variable ( recall that for the ope one has that @xmath132 ) , this expansion reads @xmath133\ \log t+ \omega^{(-)}(2,2)\ \log^2 t\bigg\}\ t^2 + \cdots\ ; .\end{gathered}\ ] ] we emphasize that only even powers of @xmath79 may appear in this expansion .
consequently , the mellin transform ( [ dict1],[dict2 ] ) becomes @xmath134(s)= \gamma(s)\ g_-(s)\\ \asymp 2\ \frac{\omega^{(-)}(0,2)}{s^3 } - \frac{\omega^{(-)}(0,1)}{s^2 } + \frac{\omega^{(-)}(0,0)}{s}\\ + 2\ \frac{\omega^{(-)}(2,2 ) } { ( s+2)^3}+ \frac{1}{12}\ \frac{6\ \omega^{(-)}(1,1)- 12\ \omega^{(-)}(2,1)- \omega^{(-)}(0,2 ) } { ( s+2)^2}\\ + \frac{1}{24}\ \frac{\omega^{(-)}(0,1)+ 6\ \omega^{(-)}(0,2)- 12\ \omega^{(-)}(1,0 ) - 12\ \omega^{(-)}(1,1)+ 24\ \omega^{(-)}(2,0)}{s+2}+ \ldots\ ; , \end{gathered}\ ] ] where the last expression corresponds to its singular expansion .
notice how the poles are located only at negative even numbers ( besides zero ) in correspondence with the even powers in the @xmath79 expansion .
again , the residues of the singular expansion are given by the coefficients of the ope in ( [ opeomega ] ) in a very specific way .
following the discussion presented in section 2 , the two singular expansions ( [ mellinthreshold ] ) and ( [ mellinope ] ) can be obtained by realizing that the @xmath6 coefficients admit an expansion for large values of @xmath8 of the form @xmath135\label{asympt2}\;,\end{aligned}\ ] ] where the powers of the logarithms @xmath136 correspond to the terms in the ope and threshold expansion which have not been yet calculated . with this expression for @xmath137 ,
it is immediate to compute the @xmath90 functions introduced in ( [ mellinpit],[integrand ] ) .
first we split the sum as @xmath138 \ n^{-s}\ .\ ] ] then , since some of the @xmath6 coefficients are known exactly , say for @xmath139 , one can actually approximate the second sum as @xmath140 n^{-s } + \delta g_{\pm}(n^*,s ) \ .\ ] ] where @xmath141 and @xmath142 \
\omega^n\ .\ ] ] in other words , up to the @xmath143 factor , @xmath144 in ( [ mellinerror ] ) is the mellin transform of the `` error '' function\ n^{-s } $ ] , as one can see in ( [ gplusminus1 ] ) .
the important point is that this expression has to be understood by analytic continuation in @xmath76 . ]
the above expression for @xmath145 makes it clear that the error in the approximation will naturally become smaller as @xmath146 gets larger , or as more terms in the expansion ( [ asympt2 ] ) for @xmath147 are included .
this is true over the whole physical region in the complex plane @xmath148 . using the expansion @xmath137 in eq .
( [ asympt2 ] ) one can readily compute @xmath149 as @xmath150 n^{-s } + \delta g_{+}(n^*,s ) \qquad , \qquad\end{gathered}\ ] ] and @xmath151 as @xmath152 n^{-s } + \delta g_{-}(n^*,s ) \qquad , \qquad\end{gathered}\ ] ] in terms of the riemann @xmath96 function and the dirichlet @xmath98 function ( recall eqs .
( [ riemann ] ) ) and their corresponding first and second derivatives . matching the two expressions ( [ mellinthreshold ] ) and ( [ mellinope ] ) to the result in eqs .
( [ gplus ] ) and ( [ gminus ] ) one readily obtains the following results for the coefficients @xmath153 and @xmath154 in ( [ asympt2 ] ) : @xmath155 @xmath156 where ( in parentheses ) we also give their approximate numerical value ( for 3 light flavors ) .
we emphasize that these @xmath157 and @xmath158 coefficients in eq .
( [ asympt2 ] ) are determined by the logarithms of the threshold expansion ( [ thresholdpi ] ) and the ope ( [ opepi ] ) in a totally unambiguous way is determined by the squared - root singularity in the threshold expansion ( [ pi2threshold ] ) , and is the only exception to this rule .
this is all in agreement with the converse mapping theorem @xcite . ] in eqs .
( [ asympt2 ] ) and ( [ asympt2values ] ) with the first 30 exact coefficients @xmath6 . ]
although the asymptotic expression ( [ asympt2 ] ) is , strictly speaking , only valid at large values of @xmath8 , it is worth realizing that ( [ asympt2 ] ) sets in very quickly , and it becomes a very efficient approximation even for relatively low values of @xmath8 . for instance , for @xmath159 the error is @xmath160 , and gets better for larger values of @xmath8 .
this could be compared to the value predicted for @xmath161 , which has a much bigger error of the order of @xmath162 .
to illustrate this fact , we plot in fig .
[ pi2omegaasomega ] the exact value of the first 30 terms for @xmath6 ( obtained from ( [ omega0=c ] ) with the exact results for the first 30 @xmath34 from ref .
@xcite ) together with the result for the asymptotic expression ( [ asympt2 ] , [ asympt2values ] ) .
this high efficiency of the asymptotic expression ( [ asympt2 ] ) will be turned into a method to predict the original taylor coefficients @xmath34 in ( [ taylorpi ] ) in the following .
one can even get an idea of the @xmath163 error in the asymptotic expression ( [ asympt2 ] ) by performing a fit to the tail of the values for @xmath34 for large @xmath8 .
although the fit is very insensitive to the sign - alternating part of ( [ asympt2 ] ) and , essentially , yields no information on the term @xmath164 , the equal - sign term @xmath163 is numerically rather well approximated by a term of the form @xmath165 . taking a @xmath166 error on the value of the coefficient ( but not on the sign ) we estimate for the error : @xmath167_{n > n^ * } \cong \left\{\begin{array}{c } + 1 \\
-0 \end{array}\right\}\ \frac{\log^{1.5 } n}{n^3 } \ ; \pm \;\ ; ( -1)^n \ \mathcal{o}\left(\frac{\log^{\ell_2 } n}{n^7}\right)\ ; .\ ] ] although the second term in the previous expression is essentially unknown , the large power @xmath168 in the denominator makes it essentially irrelevant for any reasonable value of @xmath169 .
this result will be useful for estimating the error of our calculations in what follows . given the result for @xmath102 in ( [ gplus ] ) it is straightforward to extract the residue at @xmath170 of the combination @xmath171 . according to eqs .
( [ mellinpit]-[dict2 ] ) , this residue determines the constant @xmath115 in eq .
( [ pi2threshold ] ) ( which is called @xmath116 in ref .
this prediction is interesting because it has not been possible to obtain the value for this constant by ordinary diagrammatic methods up to now .
the result yields @xmath172 + \mathcal{e}(n^ * , 1)\ ; , \label{k2}\end{aligned}\ ] ] where @xmath173 is one of the stieltjes s constants , appearing in the laurent expansion of @xmath174 around @xmath101 . in general
, one has that @xmath175\ .\ ] ] although , in principle , the expression ( [ k2 ] ) depends on the unknown term @xmath176 in eq .
( [ errorguess2 ] ) through @xmath177 in eq .
( [ error ] ) , the numerical result for @xmath116 is largely insensitive to this , for any reasonable value of the leading power @xmath169 , due to the strong suppression of the @xmath168 factor in the denominator . using eq .
( [ errorguess2 ] ) , our results in eqs .
( [ asympt2values ] ) and @xmath178 exact coefficients @xmath34 from @xcite , neglecting the second term @xmath176 , one obtains @xmath179 this result is compared to two previous determinations , using pad techniques , in table [ tablek2 ] .
@xcite & @xmath180 + @xcite & @xmath181 + ' '' ''
this work & @xmath182 + we can now double check that the contribution from the @xmath176 in eq . ( [ errorguess2 ] ) is indeed small , for any reasonable value of the power @xmath169 . in order to do this ,
let us note that analogously to the previous case , the residue at @xmath170 of the combination @xmath183 also determines the value of the constant @xmath184 ( which was called @xmath185 in ref .
the result reads @xmath186 + \mathcal{e}(n^ * , -1)\ ; .\label{h0}\end{aligned}\ ] ] however , unlike the case of @xmath116 , the result for @xmath185 has been calculated@xcite and , for 3 light flavors , it is given by @xmath187 if we neglect the term @xmath176 in eq .
( [ errorguess2 ] ) , we can compute the value for the expression ( [ h0 ] ) with the help of our results in eqs .
( [ asympt2values ] ) and the @xmath178 exact coefficients @xmath34 ( @xmath6 ) .
the result obtained is @xmath188= -0.5856\ ^{+ 0.0001}_{-0.0000 } \ $ ] , in agreement with ( [ h02 ] ) . in eq .
( [ prediction2 ] ) compared with the exact @xmath34 , for different values of @xmath189 , @xmath190 and @xmath191 , as explained in the text . also shown is the direct asymptotic expression for @xmath34 obtained in ref .
@xcite ( labeled as `` asymp'').,scaledwidth=90.0% ] expression ( [ c = omega0 ] ) gives us the exact dictionary to translate the values of the coefficients @xmath6 into those for @xmath34 .
exact knowledge of a few @xmath34 ( let us say for @xmath139 ) together with the asymptotic expression @xmath137 , allows us then to make a _ prediction _ for the rest of coefficients for @xmath192 , as we will now describe . splitting the sum in (
[ c = omega0 ] ) between @xmath193 and @xmath194 and using the exact value of @xmath195 for the first part and the asymptotic expression and @xmath196 ( [ asympt2 ] ) for the second part , one obtains the following prediction @xmath197 : @xmath198\;,\end{gathered}\ ] ] the values obtained for @xmath197 can now be compared , e.g. , with the result of the direct asymptotic expression for @xmath34 obtained in ref .
@xcite ( black curve in fig .
[ errorc2n ] ) .
as one can see in this figure , already for @xmath189 ( i.e. only assuming that the two coefficients @xmath199 and @xmath200 are exactly known ) , one achieves a better approximation than that obtained directly in terms of the coefficients @xmath34 , , for which there seems to be a fortuitous cancelation . ] with errors which are of the order of 6 per mil , at worst ( blue curve ) . on this figure
we also show the curves when @xmath190 ( and @xmath191 ) , i.e. when 3 ( respectively 4 ) coefficients @xmath34 are exactly known . in these cases , the errors are even smaller ( green ( respectively red ) curves ) .
this shows that eq .
( [ prediction2 ] ) together with the asymptotic expression ( [ asympt2 ] , [ asympt2values ] ) is a good strategy to predict the value of the coefficients @xmath34 .
this fact will be relevant in the case of the function @xmath201 . starting from eq .
( [ taylorpiw ] ) , we can again rearrange the series by singling out the first @xmath146 coefficients , again assumed to be exactly known , and write @xmath204\omega^n + \mathcal{e}(n^*,\omega ) \;,\ ] ] where @xmath205\omega^n \;,\ ] ] is the `` error function '' already defined in eq .
( [ error ] ) . the first term in eq .
( [ approxpi2v1 ] ) can be explicitly summed using the fact that is nothing else than the usual polylogarithm @xmath206 .
here we write in this slightly unconventional way because we need to differentiate with respect to the variable @xmath76 .
we also use the notation @xmath207 to signify @xmath208 . ]
@xmath209 where the polylogarithm may be defined in the @xmath7 complex plane as @xmath210 consequently , we find the following expression for @xmath203 , @xmath211\omega^n + \mathcal{e}(n^*,\omega ) \ ; , \end{aligned}\ ] ] where the coefficients @xmath153 and @xmath154 are given in ( [ asympt2values ] ) . neglecting the error function @xmath212 ( an approximation which gets better as @xmath146 grows larger , and also as many more terms in the expansion ( [ asympt2 ] )
are known ) , our expression for @xmath213 is given by a unique combination of polylogarithms ( and derivatives thereof ) plus a known polynomial of degree @xmath146 in @xmath7 .
this expression constitutes a resummation of the taylor expansion ( [ taylorpiw ] ) thanks to the information supplied by the threshold expansion and the ope , as seen in eqs .
( [ thresholdpi]),([opepi ] ) and ( [ asympt2values ] ) .
notice that the knowledge of further terms in either expansion can only lead to the determination of further @xmath157 s and @xmath158 s coefficients in @xmath137 , leaving unaltered the coefficients already determined .
therefore , the approximation is systematic and can be improved upon by either knowing more @xmath34 coefficients exactly ( i.e. increasing @xmath146 ) or by knowing more terms from the threshold expansion or the ope ( i.e. by knowing more @xmath157 s and @xmath158 s in @xmath137 ) .
this should be clear from the fact that the expression for the error function @xmath212 in eq .
( [ errorfunction ] ) naturally becomes smaller in these two cases , for any @xmath148 in the complex plane . in order to further illustrate this point , in the appendix
we show the result of our resummation for the spectral function and compare it to the exact result , in the particular case of the function @xmath214 in eq .
( [ pi ] ) , where this result is fully known . , in terms of the quark velocity @xmath215 .
the dashed curves correspond to consecutive approximations from the threshold expansion ( @xmath216 ) and from the ope ( @xmath217).,scaledwidth=85.0% ] undoing the change of variables ( [ conformal ] ) , and using the expression for @xmath218 in ( [ result2omega ] ) , one immediately obtains @xmath219 as an representation for the function @xmath220 in the whole @xmath22 complex plane .
in particular , the analytic continuation of the polylogarithms and the polynomial in @xmath7 yield an approximation to its branch cut and the corresponding spectral function . in principle
, we could easily give the explicitly analytic result for this spectral function from the expression ( [ result2omega ] , [ piapprox2 ] ) but we refrain from doing so because the result is not particularly illuminating .
alternatively , we show in fig .
[ pi2impartn3 ] the result for this spectral function as a function of the quark velocity @xmath221 for @xmath190 , together with the corresponding approximations at threshold ( @xmath216 ) and the ope ( @xmath217 ) ( dotted curves ) . in this plot ,
the error band is estimated with the help of the error function @xmath222 in ( [ result2omega ] ) and the educated guess ( [ errorguess2 ] ) .
even for @xmath190 the errors are so small that they can not be seen on the plot , becoming even smaller at larger values of @xmath146 .
this plot may be compared to the one in fig .
one could of course use all the known @xmath34 coefficients , i.e. take @xmath178 in our expressions .
however , at the current level of precision , this is not yet necessary @xcite .
we can now apply the same method exposed in the previous section for the @xmath112 function , to treat the function @xmath111 . for later convenience ,
let us first reproduce here the coefficients of the taylor , ope and threshold expansions listed in ref .
@xcite for the vector vacuum polarization function at @xmath20 ( for 3 light flavors ) , @xmath223 using the coefficients of these expansions , a straightforward rerun of the method leads to the following result for the asymptotic expression of the corresponding coefficients @xmath224 in this case : @xmath225\;,\end{aligned}\ ] ] with @xmath226 in eqs .
( [ omegaas3 ] ) and ( [ values3 ] ) ( blue crosses ) and the exact value for the first three coefficients @xmath6 , @xmath227 ( red circles ) . ,
width=480 ] this expression for @xmath137 can now be compared with the exact numerical result for @xmath92 which can be obtained from the exact value of the first 3 coefficients @xmath34 , @xmath227 , calculated in ref .
@xcite , and our previous formula ( [ omega0=c ] ) .
the result is shown in fig .
[ omega3plot ] . as one can see , the asymptotic form ( [ omegaas3 ] , [ values3 ] ) sets in very quickly , as it also happened in the previous case of the function @xmath112 ( see fig .
[ pi2omegaasomega ] ) , reproducing the exact values for @xmath228 and @xmath229 within only a few per cent error .
notice that the error for @xmath230 would be much larger @xmath162 , as it is clearly seen in fig .
[ omega3plot ] . based on this success for the `` prediction '' of @xmath228 and @xmath229
, we can make a very rough guess for the error of the asymptotic expression ( [ omegaas3 ] ) in the form : @xmath231_{n > n^*(\sim 2 - 3 ) } \cong \pm \
15\ \frac{\log^{3 } n}{n^2 } \ ; \pm \;\ ; ( -1)^n \ \mathcal{o}\left(\frac{\log^{\ell_2 } n}{n^7}\right)\ ; .\ ] ] at present we know nothing about the sign - alternating term @xmath232 in the expression above although , fortunately , one is protected by the high power in the denominator , i.e. @xmath168 , for any reasonable value of the @xmath233 power .
this also happened in the case of the function @xmath112 ( see eq .
( [ errorguess2 ] ) ) .
therefore , the error coming from the first term in ( [ erroromega3 ] ) and , moreover , its slow fall - off with @xmath8 , i.e. @xmath234 , will now be the main source of error in all our estimates .
it is important to realize that the error ( [ erroromega3 ] ) can only be reliably reduced once the terms of @xmath235 in eq .
( [ omegaas3 ] ) are computed , which in turn requires the calculation of the terms @xmath236 in the threshold expansion ( the second of the eqs .
( [ expansion3 ] ) ) .
regretfully , this information is not available at present .
alternatively , one could also compute more @xmath34 coefficients from the taylor expansion , in order to increase the value of @xmath146 .
however , this second possibility does not seem so promising due to the rather slow @xmath234 fall - off with @xmath8 shown by the non sign - alternating part of the expression in eqs .
( [ omegaas3 ] ) and ( [ erroromega3 ] ) . at any rate , using the expression for @xmath137 in ( [ omegaas3],[values3 ] ) , the result for the constant @xmath237 in eq . ( [ expansion3 ] ) in eq .
( 19 ) of ref .
] , akin to the constant @xmath116 in the section 3.1 , now becomes @xmath238 + \mathcal{e}(n^*,1)\quad , \end{aligned}\ ] ] where @xmath239 are the stieltjes s constants defined in eq .
( [ stieltjes ] ) . using again the 3 exactly known coefficients @xmath6 ( @xmath227 ) from @xcite ( i.e. @xmath190 ) and our estimate for the error in the coefficients ( [ erroromega3 ] ) to compute the error function @xmath240 defined in ( [ error ] ) ,
we obtain : @xmath241 although our central value almost coincides with that in ref .
@xcite , our error is @xmath242 8 times larger than that quoted in this reference and , in particular , can not exclude a negative value for @xmath243 , as it was found in ref .
the situation is summarized in table [ tablek3 ] .
it is now immediate to assess the impact of a reduction of the systematic error ( [ erroromega3 ] ) .
for instance , if this error were reduced to @xmath244 one would get an error in @xmath237 , @xmath245 , whereas if the error could be pushed down to @xmath246 , the error would be @xmath247 ; a dramatic improvement .
@xcite & @xmath248 + @xcite & @xmath249 + ' '' '' this work & @xmath250 + as a cross - check of our method , we can also calculate the value of the first constant @xmath251 appearing in the ope @xcite .
we obtain @xmath252(-1)^n + \mathcal{e}(n^*,-1 ) \quad .\end{aligned}\ ] ] using the values obtained in ( [ values3 ] ) , our estimate ( [ erroromega3 ] ) and @xmath190 , we obtain @xmath253 while the exact value is @xmath254 @xcite . the reason why this estimate is much better than ( [ k3value ] )
is because @xmath237 is more sensitive to the non sign - alternating component of the error ( [ erroromega3 ] ) , which is large .
further constants , such as @xmath255 and @xmath256 in ( [ expansion3 ] ) , are more sensitive to this component of the error , due to the extra powers of @xmath8 in the sums , resulting in a larger total error and we refrain from giving their result . concerning our prediction for the coefficients @xmath197
, we can use the exact value for the first three @xmath34 , @xmath227 , again from @xcite , and our result in ( [ prediction2 ] ) with the expression for @xmath137 from ( [ omegaas3 ] ) and the values ( [ values3 ] ) to obtain the results listed on table [ coeff ] .
of difference between our normalization and that in ref . @xcite .
] we find agreement with the numbers quoted in @xcite , although our central values are systematically lower and the errors are larger .
.table of comparison between the @xmath197 from this work and the @xmath34 in @xcite . [
cols="^,^,^,^,^",options="header " , ] finally , we can also give the expression for @xmath257 obtained by summing the coefficients ( [ omegaas3],[values3 ] ) , in complete analogy to what we did in eq . ( [ approxpi2v1 ] ) .
the result is @xmath258\omega^n + \mathcal{e}(n^*,\omega ) \ ; .\end{aligned}\ ] ] again , we plot in fig .
[ spectral3 ] the result for the spectral function which comes from the imaginary part of ( [ result3omega ] ) .
this figure may be compared to the akin figures in ref .
@xcite ( fig . 5 ) and ref .
@xcite ( fig .
2 , first panel ) . ) in terms of the quark velocity , @xmath259 , together with the threshold approximation ( @xmath216 ) ( in green ) and the ope ( @xmath217 ) ( in red).,width=576 ]
given an analytic function with a cut located at @xmath260 , we have presented a method for reconstructing the full function from its taylor expansion around the origin , i.e. @xmath261 , eq . ( [ taylorpi ] ) , its threshold expansion for @xmath262 , eq . ( [ thresholdpi ] ) , and its ope series for @xmath263 , eq .
( [ opepi ] ) .
we have exemplified this method with the case of the perturbative vacuum polarization function of a quark , but we think that it may be useful in other contexts as well .
the method uses the conformal variable @xmath7 , eq .
( [ conformal ] ) , in terms of which the whole cut complex plane is mapped onto a disk of unit radius ( with the cut running at the boundary , see fig . [ omegaplan ] ) , and consists in realizing that , in this variable , the taylor coefficients @xmath6 , eq .
( [ taylorpiw ] ) , admit an asymptotic expansion in inverse powers of @xmath8 , for large @xmath8 , with a sign - alternating component and an equal - sign component .
the coefficients of the equal - sign component , which we call @xmath108 in eq .
( [ defomega ] ) , are essentially determined by the logarithms of the threshold expansion , whereas the coefficients of the sign - alternating component , which we call @xmath109 in eq .
( [ defomega ] ) , are determined by the logarithms of the ope .
since logarithms are controlled by the renormalization group , it is not inconceivable that one may also link this asymptotic behavior in @xmath8 to the renormalization group in the future .
we have found that the precise dictionary , eqs .
( [ dict1],[dict2 ] ) , is most easily established by means of the use of the mellin transform and the converse mapping theorem exposed in ref . @xcite and references therein .
once an asymptotic expression for the coefficients @xmath137 is obtained , the power series can be explicitly summed using this asymptotic expression starting from a certain @xmath146 onwards . below this @xmath146
, the exact values for the coefficients must be used .
the precision of the result of course depends on how precise the value of @xmath137 is for @xmath264 .
interestingly , we have found that the asymptotic series @xmath137 sets in very quickly , giving quite accurate results already for @xmath265 .
the resummed function in terms of the conformal variable @xmath7 can then be expressed as a combination of polylogarithms and their derivatives plus a polynomial , with known coefficients ( see , e.g. eqs .
( [ result2omega ] ) , ( [ result3omega ] ) ; and the appendix ) .
the method is analytic and it is possible to know how our results depend on our ignorance due to uncalculated terms in the taylor , ope and threshold expansion ; which has given us a way to estimate the associated systematic error ( [ error ] ) .
we have made some numerical comparisons with the results obtained via the pad method @xcite , and our results agree with them , although our errors are usually larger .
nevertheless , we think it would be very interesting to be able to make a more thorough comparison including , in particular , the result for the full spectral function .
furthermore , we plan to use our method to study other channels as well @xcite .
regarding the @xmath20 vacuum polarization function , @xmath201 , we find that the calculation of all the terms @xmath266 from the threshold expansion for @xmath267 , or at least the one with the highest power @xmath268 , would result in a substantial reduction of the associated theoretical error in all the predictions for this function .
we thank p. masjuan for his collaboration at an early stage of this work and a. pineda for discussions .
we also thank s. friot , m. golterman , m. jamin and e. de rafael for comments on the manuscript .
this work has been supported by cicyt - feder - fpa2008 - 01430 , sgr2005 - 00916 , the spanish consolider - ingenio 2010 program cpan ( csd2007 - 00042 ) and by the eu contract no .
mrtn - ct-2006 - 035482 , `` flavianet '' .
in this appendix we will use a known function to further illustrate our method . for this purpose
, we will choose the @xmath15 result , @xmath214 , which reads explicitly @xmath269\ ] ] where @xmath270 and @xmath271 using the method we describe in the text , the ope and threshold expansions for this function determine that the asymptotic expansion of the associated coefficients @xmath6 in the taylor expansion of @xmath272 in the conformal variable @xmath7 are given by @xmath273\ ; .\end{aligned}\ ] ] amusingly , there are only sign - alternating coefficients , i.e. @xmath154 , but no @xmath153 coefficient .
this feature is in agreement with the fact that the threshold expansion of @xmath214 does not have any @xmath274 term , so all the terms in @xmath137 are given by the @xmath275 terms of the ope . as also happens for @xmath202 and @xmath276 , the asymptotic expansion @xmath137 sets in very quickly . in fig .
[ toymodelomega ] we compare the exact values for the coefficients @xmath6 and this asymptotic expression . as one
can see , already for @xmath277 the error is only a few percent , becoming smaller for larger @xmath8 .
@xmath280\ \omega^n{\nonumber}\\ & = & \frac{1}{2\pi^2}\log(1+\omega)+\frac{9}{32\pi^2}\li(5,-\omega)+ \sum_{n=1}^{n^*}\left [ \omega(n)-\omega^{as}(n)\right ] \
\omega^n\qquad , { \nonumber}\end{aligned}\ ] ] which , as usual , depends on how many coefficients @xmath6 we assume to be exactly known ( @xmath146 ) . in fig .
[ toymodelspectrum ] we show how this expression approximates the exact function over the whole complex plane by comparing the spectral function obtained from this approximation to the exact spectral function of the function @xmath214 , for different values of @xmath146 . ) , it is clear that approximating the spectral function is equivalent to approximating the full function in the whole complex plane.]as we can see on this plot , even for @xmath189 ( i.e. assuming only that the first two @xmath6 coefficients are exactly known ) we reproduce the exact result with a relative error at the per mil level ( or smaller ) , over the whole range in velocity @xmath281 . for larger values of @xmath146
the errors are , of course , even smaller .
j. h. kuhn , m. steinhauser and c. sturm , nucl .
b * 778 * , 192 ( 2007 ) [ arxiv : hep - ph/0702103 ] .
k. g. chetyrkin , j. h. kuhn , a. maier , p. maierhofer , p. marquard , m. steinhauser and c. sturm , phys .
d * 80 * , 074010 ( 2009 ) [ arxiv:0907.2110 [ hep - ph ] ] .
j. h. kuhn , arxiv:1001.5173 [ unknown ] .
i. allison _ et al . _
[ hpqcd collaboration ] , phys .
d * 78 * , 054513 ( 2008 ) [ arxiv:0805.2999 [ hep - lat ] ] .
a. o. g. kallen and a. sabry , kong .
sel . mat .
fys . med .
* 29n17 * ( 1955 ) 1 .
r. boughezal , m. czakon and t. schutzmeier , nucl .
proc . suppl .
* 160 * , 160 ( 2006 ) [ arxiv : hep - ph/0607141 ] .
a. maier , p. maierhofer and p. marquard , nucl .
b * 797 * , 218 ( 2008 ) [ arxiv:0711.2636 [ hep - ph ] ] .
k. g. chetyrkin and j. h. kuhn , nucl .
b * 432 * , 337 ( 1994 ) [ arxiv : hep - ph/9406299 ] ; k. g. chetyrkin , j. h. kuhn and m. steinhauser , phys .
b * 371 * , 93 ( 1996 ) [ arxiv : hep - ph/9511430 ] .
k. g. chetyrkin , j. h. kuhn and m. steinhauser , nucl .
b * 482 * , 213 ( 1996 ) [ arxiv : hep - ph/9606230 ] .
k. g. chetyrkin , r. harlander , j. h. kuhn and m. steinhauser , nucl .
b * 503 * , 339 ( 1997 ) [ arxiv : hep - ph/9704222 ] .
a. czarnecki and k. melnikov , phys .
* 80 * , 2531 ( 1998 ) [ arxiv : hep - ph/9712222 ] .
y. kiyo , a. maier , p. maierhofer and p. marquard , nucl .
b * 823 * , 269 ( 2009 ) [ arxiv:0907.2120 [ hep - ph ] ] .
d. j. broadhurst , j. fleischer and o. v. tarasov , z. phys .
c * 60 * ( 1993 ) 287 [ arxiv : hep - ph/9304303 ] ; j. fleischer and o. v. tarasov , z. phys .
c * 64 * ( 1994 ) 413 [ arxiv : hep - ph/9403230 ] ; p. a. baikov and d. j. broadhurst , arxiv : hep - ph/9504398 ; k. g. chetyrkin , j. h. kuhn and m. steinhauser , phys .
b * 371 * ( 1996 ) 93 [ arxiv : hep - ph/9511430 ] ; k. g. chetyrkin , j. h. kuhn and m. steinhauser , nucl .
b * 482 * ( 1996 ) 213 [ arxiv : hep - ph/9606230 ] .
a. h. hoang , v. mateu and s. mohammad zebarjad , nucl .
b * 813 * , 349 ( 2009 ) [ arxiv:0807.4173 [ hep - ph ] ] .
flajolet , x. gourdon and ph .
dumas , theoretical computer science * 144 * ( 1995 ) 3 . | with the help of the mellin - barnes transform , we show how to simultaneously resum the expansion of a heavy - quark correlator around @xmath0 ( low - energy ) , @xmath1 ( threshold , where @xmath2 is the quark mass ) and @xmath3 ( high - energy ) in a systematic way .
we exemplify the method for the perturbative vector correlator at @xmath4 and @xmath5 .
we show that the coefficients , @xmath6 , of the taylor expansion of the vacuum polarization function in terms of the conformal variable @xmath7 admit , for large @xmath8 , an expansion in powers of @xmath9 ( up to logarithms of @xmath8 ) that we can calculate exactly .
this large-@xmath8 expansion has a sign - alternating component given by the logarithms of the ope , and a fixed - sign component given by the logarithms of the threshold expansion in the external momentum @xmath10 .
* resummation of threshold , low- and high - energy + expansions for heavy - quark correlators * + * david greynat@xmath11 * and * santiago peris@xmath12 * + @xmath11ifae , @xmath13grup de fsica terica , dept . de fisica + univ .
autnoma de barcelona , 08193 barcelona , spain . + |
a skyrmion , first introduced in high - energy physics in the description of baryons,@xcite is a topologically non - trivial configuration of a continuous field @xmath2 , which is localized , in the sense that its gradient @xmath3 is significant only in a finite spatial region @xmath4 . far away from @xmath4 , the field @xmath2
approaches a uniform configuration , which is conveniently described by a boundary condition at spatial infinity.@xcite the euclidean action @xmath5 $ ] usually has the form of non - linear sigma model ( nlsm ) , with the lower bounds of energy being guaranteed by the bogomolnyi prasad sommerfield inequality .
@xcite such lower bound is proportional to @xmath6 the absolute value of an integer `` topological charge '' @xmath7 , which classifies the field @xmath2 as a continuous mapping . in the work of skyrme , @xcite @xmath2 maps the ( compactified ) 3 + 1 dimensional space - time to the manifold of lie group @xmath8 ( which is isomorphic to @xmath9 , the 3-sphere ) , and the third homotopy group @xmath10 provides the topological charge @xmath7 , which was identified as particle number in that work .
after its introduction in high - energy physics , skyrmion physics regained interest in the context of condensed - matter physics as a topological object in a two - dimensional ferromagnetic state .
they have been discussed in quantum hall systems,@xcite and in chiral magnets@xcite and magnetic thin films.@xcite while skyrmions have been experimentally viewed in scanning - tunnelling microscopy and spectroscopy ( stm / sts ) in the latter two systems,@xcite skyrmions in quantum hall systems have mainly been identified through their magnetic properties in nuclear - magnetic - resonance experiments,@xcite and a spectroscopic identification is yet outstanding .
furthermore , skyrmions have been used to describe special configuration of the superconducting order parameter in superconductors.@xcite the magnetization or the superconducting order parameter , denoted as @xmath2 , usually maps the ( compactified ) two - dimensional plane to the bloch sphere @xmath11 .
the second homotopy group @xmath12 provides the topological charge @xmath7 , which is the degree of the mapping between two 2-spheres .
a more familiar case is the vortex configuration of the complex scalar field @xmath2 on two - dimensional plane , which is punctured by the vortex core .
the phase of the complex scalar field lives in a ring @xmath13 .
then the homotopy group @xmath14 provides us the topological charge @xmath7 , known as winding number .
some more exotic cases appeal to the second homotopy group of@xcite @xmath15 that allows for establish of skyrmions in @xmath16-component systems , such as in refs .
[ ] for @xmath17 and [ ] for @xmath18 , or , to the homotopy group @xmath19 in ref .
[ ] . in the present study , we choose a @xmath0 field on two - dimensional plane to model the static configuration of qh skyrmions in graphene monolayer .
the distinguished feature of quantum hall ( qh ) skyrmion is that it carries _ electric _ charge of @xmath20 , where @xmath7 is its _ topological _ charge and @xmath21 is the charge of an electron .
moreover , such equality holds locally , i.e. the electric charge is smeared in the texture of the qh skyrmion , the excess charge density @xmath22 in a qh skyrmion is equal@xcite to @xmath23 , where @xmath24 characterizes the spatial variation of the texture in the qh skyrmion , and the integral @xmath25 reproduces its topological charge @xmath7 . as a consequence of the smeared charge density , a qh skyrmion has a lower energy in the @xmath1 landau level ( ll ) than the quasiparticle of the same charge , therefore reduces the charge transport gap .
a lattice of qh skyrmions also has lower energy than the corresponding wigner crystal.@xcite we finally notice that , as a consequence of the intimate relation between the topological and the electric charge , a skyrmion is a fermionic excitation in qh systems , in contrast to most of the aforementioned magnetic systems . in this work
, we study the qh skyrmions in the @xmath1 ll in graphene monolayer .
besides the robust quantum hall effects in graphene in a wide range of temperature@xcite and magnetic field,@xcite there are several advantages to choose graphene as the host system for qh skyrmions .
first , the dirac valley of electron states can be described by a pseudospin degree of freedom , and in the presence of strong perpendicular magnetic field the coulomb interaction is approximately symmetric@xcite under global @xmath26 transformations in the spin@xmath27valley space .
in contrast , the layer pseudospin in double - layer qh system does not enjoy such a symmetric coulomb interaction in the spin@xmath27valley space , because the layer separation always breaks the pseudospin symmetry of the coulomb interaction . in our approach ,
the @xmath26 spin - valley symmetry is explicitly broken by terms at lower energy scales ( compared to the coulomb energy ) , namely the zeeman coupling that couples directly to the spin , and the valley pseudospin anisotropy . as a second advantage of the graphene monolayer ,
one may point out that the valley degree of freedom coincides with the sublattice degree of freedom for the electron states in the @xmath1 ll .
such a coincidence provides a convenient way to probe the @xmath28-component of the valley pseudospin directly it can be read off from the sublattice occupation , thus allowing a direct imaging of the pseudospin texture via stm / sts experiments .
moreover , the spin texture in graphene under qh conditions can also be imaged by spin - resolved stm / sts experiments .
the combination of the lattice - resolved images for spin and pseudospin textures thus allows for an identification of the various skyrmion types in graphene .
the fact that graphene is naturally a surface system renders these states spectroscopically more accessible than qh systems in semiconductor heterostructures , where the two - dimensional electron system appears at the interface between two types of semiconductors . for stm / sts experiments in semiconductor heterostructures , two - dimensional electron systems have been realized on an n - doped insb surface@xcite albeit with a mobility that does not reach that of the more common gaas hetero - structures or graphene .
our work completes our previously obtained phase diagram for qh skyrmions in graphene@xcite in several respects .
apart from a full description of the fm states one encounters at @xmath29 ( and @xmath30 by particle - hole symmetry ) , we provide here a detailed characterization of skyrmions with entanglement .
in addition to a pure entanglement skyrmion , one obtains an exotic type of _ deflated _ pseudospin skyrmions with partial entanglement .
we characterize all encountered skyrmion types with the help of three different bloch spheres that describe spin , valley pseudospin and entanglement , respectively .
furthermore , all ferromagnetic and skyrmion states are analysed from their appearance in lattice - resolved density plots that may use as a guideline in an experimental stm / sts investigation .
finally , we discuss the scaling relations of the skyrmion size and energy close to the transitions between different underlying fm states
while a critical behavior is obtained in the case of a symmetry restoration at the transition , we find that the scaling is truncated in the other cases where a ( subordinate ) symmetry is not fully restored .
these scaling relations are then discussed in the context of further modifications of the anisotropy parameters that we vary to obtain the skyrmion phase diagram , e.g. in the case of a tilted sample .
the paper is organized as follows . in sec .
[ sec : quantum - hall - ferromagnetic - state ] we propose an anisotropic energy for the ferromagnetic state in the @xmath1 ll at quarter filling , and discuss the @xmath0 representation of the qh ferromagnetic states .
there are four types of ferromagnetic ground states for the anisotropic hamiltonian .
each type is described by a @xmath0 spinor that contains six angular variables and visualized on the spin , pseudospin and entanglement bloch spheres .
we also visualize them on the honeycomb lattice . in sec .
[ sec : cp3-skyrmion - of - charge - q-1 ] we discuss the energy functional and ansatz for the @xmath0 skyrmion on top of the qh ferromagnetic states , and present the skyrmion type diagram , which is obtained by minimization of the energy functional with the ansatz and under various input parameters of the model
. the detailed analysis of the minimization result is presented in sec .
[ sec:4-types - skyrmion ] .
we visualize the skyrmion @xmath0-field on three bloch spheres in a same manner as for the qh ferromagnetic states .
these visualizations explicitly demonstrate that the skyrmions can be viewed as a wrapping of the @xmath31-plane on bloch spheres . besides , the @xmath0-skyrmion is also visualized on the honeycomb lattice , in order to show the difference between two types of skyrmions .
the lattice - resolved profiles of the @xmath0-skyrmion serve as references for the stm / sts imaging of spin - pseudospin textures in graphene monolayer under qh conditions .
the qualitative discussion on the @xmath0-skyrmions are followed by quantitative discussions in sec .
[ sec : size_and_energy ] , where we present the dependence of size and energy of the @xmath0-skyrmions on the input parameters of the model . finally in sec .
[ sec : modification - of - the - anisotropy - energy ] , we discuss three ways to modify the anisotropic energy of the @xmath0-skyrmions .
we demonstrate how the type of a @xmath0-skyrmion is altered by these modifications .
these modifications are relevant in certain experimental regimes .
a multicomponent quantum hall system is characterized by the @xmath32-fold degeneracy of the landau levels which originates from internal degrees of freedom of 2d electrons in a strong magnetic field .
for example , when the zeeman energy is much smaller than the coulomb interaction , one may first neglect the zeeman coupling of the electron spin to the applied magnetic field and consider spin as an internal degree of freedom which yields a two - fold degeneracy of the landau levels . at half - filling of such ll ( i.e. integer filling of landau sublevel ) ,
the major cause for the complete spin polarization in the ground state is the coulomb interaction , which is larger than the zeeman energy and prefers a totally asymmetric orbital part of the many - particle wave function , so that the electrons are separated as far as possible , and the spinor part is totally symmetric .
the zeeman coupling at a smaller energy scale then conspires with the coulomb energy and orients the polarized spins along the applied magnetic field . in this way ,
the electrons in a single ll form a _ quantum hall ferromagnet _ ( qhfm ) . in graphene monolayer
, the internal degrees of freedom of the landau levels are doubled by the valley degeneracy , which is described by a _ pseudospin _
analogous to the electron spin .
the coulomb interaction has approximate @xmath26 symmetry,@xcite which is broken by spin / pseudospin anisotropic interactions at smaller energy scale . at quarter filling , i.e. when one of the four spin - valley sublevels is completely filled ( @xmath29 for electron filling ) or completely empty ( @xmath33 for hole filling ) , the ground state @xmath34 for the electron system is the @xmath26 quantum hall ferromagnetic ( qhfm ) state@xcite @xmath35 where @xmath36 creates an electron in the @xmath1 ll with spin @xmath37 ( @xmath38 ) and pseudospin @xmath39 ( @xmath40 ) .
the quantum number @xmath41 is the landau orbit , which characterizes the ll degeneracy .
it is related to the underlying translation invariance
while this invariance is , strictly speaking , broken by the position - dependent vector potential , the physical magnetic field is constant in the entire plane .
the kinetic energy of an electron in this field does therefore not depend on the center of its quantized cyclotron motion .
the landau orbit @xmath41 is precisely the quantum number associated with the position of this center . in the case of one electron per landau orbit , which we consider here , the coefficients @xmath42
satisfy @xmath43 . to model electron states with spatially homogeneous spin - pseudospin magnetization at a length - scale larger than @xmath44
, we assume that @xmath45 do not carry landau orbit index @xmath41 . since @xmath46 and @xmath47 ( @xmath48 is a real constant ) correspond to the same quantum state , the coefficients in eq .
( [ eq : qhfm ] ) can be uniquely represented by a cp@xmath49-spinor @xmath50 , which is normalized @xmath51 and is equivalent to @xmath52 .
( see appendix .
[ subsec : cp3-spinor - and - nlsm ] ) the meaning of the four components in the cp@xmath49-spinor is inherited from @xmath46 . in this section ,
based on a general proposal of valley pseudospin anisotropy , we discuss different types of qhfm ground states at quarter filling ( @xmath29 for electron filling and @xmath33 for hole filling ) of the four - fold degenerate @xmath1 ll in graphene monolayer .
we propose the anisotropic energy in sec .
[ subsec : anisotropic - energy ] , and then discuss the parametrization of cp@xmath49-spinor in sec .
[ subsec : parametrization ] .
the four types of qhfm ground state are discussed in sec .
[ subsec : qhfm - ground - states ] .
we schematically visualize the qhfm states on the honeycomb lattice of graphene in sec .
[ subsec : qhfm - visualization ] .
the electrons are restricted to the @xmath1 ll in our problem and hence the kinetic energy is quenched and set to a constant .
due to the @xmath26 symmetry of the interaction hamiltonian@xcite
@xmath53 the coulomb energy @xmath54=\left\langle f\right|h_{\rm c}\left|f\right\rangle$ ] does not depend on the qhfm state @xmath46 .
the @xmath26 symmetry is explicitly broken by the spin - pseudospin anisotropies @xcite @xmath55 where @xmath56 and @xmath57 are spin and pseudospin density operators .
the first terms reflect a pseudospin anisotropy that can be generated , e.g. by short - range interactions of hubbard type@xcite ( @xmath58 and @xmath59 ) or an out - of - plane@xcite ( also contributing to @xmath59 ) and in - plane kekul - type @xcite lattice deformation due to electron - phonon couplings .
the coefficients @xmath60 have the dimension of energy , and @xmath61 are proportional to the perpendicular component @xmath62 of the applied magnetic field , whereas @xmath63 is proportional to the total applied magnetic field @xmath64 .
their numerical values are estimated as @xmath65k$ ] , @xmath66k$ ] , @xmath67k$ ] and @xmath68k$ ] respectively in ref .
@xmath69 explicitly breaks the @xmath26 symmetry down to u@xmath70u@xmath71u@xmath72 .
while the u@xmath73 and the u@xmath74 symmetries reflect rotations around the @xmath28-quantization axes of the spin and pseudospin , respectively , @xmath75 indicates that the two orientations @xmath28 and @xmath76 of the pseudospin are equivalent .
they are not equivalent in the spin channel due to the zeeman coupling .
the u@xmath77 symmetry is more subtle and related to spin - pseudospin entanglement,@xcite as we discuss in more detail below .
the system energy is determined by minimization of the anisotropic energy @xmath78=\left\langle f\right|h_{\rm a}\left|f\right\rangle$ ] , which we propose to be the following form : @xmath79 = a_{0}\frac{\delta_{\rm z}}{2}\left\ { u_{\perp}\left(m_{\rm px}^{2}+m_{\rm py}^{2}\right)+u_{\rm z}m_{\rm pz}^{2}-m_{\rm sz } \right\}.\ ] ] where @xmath80 is a dimensionless area to ensure an extensive energy , @xmath81 , and the dimensionless parameters @xmath82 and @xmath83 characterize the pseudospin anisotropy . in the present work , @xmath82 and @xmath83
are treated as control parameters that generate the different phases even if they may be more difficult to vary in a typical situation . in the above equation
, we denote the spin and pseudospin magnetizations as @xmath84 respectively .
they can be expanded explicitly in the components of the @xmath85-spinor @xmath86 .
for instance , the @xmath28-component of the spin magnetization @xmath87 is @xmath88 and the @xmath28-component of the pseudospin magnetization @xmath89 is @xmath90 where we use the labels @xmath91 , @xmath92 , @xmath93 and @xmath94 for the index @xmath95 of the components of @xmath86 to emphasize their meaning .
it is essential for later discussions to use the schmidt decomposition to parametrize a general cp@xmath49-spinor @xmath96 with six real parameters , namely @xmath97 $ ] and @xmath98:@xcite @xmath99 where @xmath100 and @xmath101 , @xmath102 are the pauli matrices , @xmath103 is the unit vector for the direction of magnetization
. the subscript @xmath104 ( @xmath105 ) stands for spin ( pseudospin ) .
the pauli matrices for spin and pseudospin are designed as @xmath106 and @xmath107 respectively , in agreement with the earlier convention for the cp@xmath49-spinor @xmath86 . in the present section , where we discuss qhfm states , the real parameters @xmath108 , and @xmath109 are constant in space , i.e. independent of the landau orbit @xmath41 , while we consider an explicit position dependence in the following section on skyrmions . to simplify notations ,
we therefore omit here to write explicitly the argument @xmath110 in the spinors and parameters . under the parametrization eq .
( [ eq : parametrizationz ] ) , the spin and the pseudospin magnetization for the cp@xmath49-spinor @xmath96 are @xmath111 respectively .
the meaning of @xmath112 and @xmath113 in the parametrization eq .
( [ eq : parametrizationz ] ) is evident from the above equations
they are the polar and azimuthal angle of the spin and pseudospin magnetizations .
as we discuss in detail later , the parameter @xmath114 enriches the types of fm ground state of the system .
it can be understood in terms of `` entanglement''.@xcite it is possible to rewrite the state @xmath96 into a direct product of spinors for electron spin and and valley pseudospin _ only _ when @xmath115 or @xmath116 . in this case
the spin and pseudospin are _
unentangled_. otherwise , @xmath96 is a superposition of two `` product states '' carrying opposite spin and pseudospin .
consequently , as indicated in eq .
( [ eq : si_y ] ) and eq . ( [ eq : ti_y ] ) , the spin and pseudospin magnetizations have magnitudes smaller than @xmath117 . remarkably , these magnitudes are equal to @xmath118 , because in a @xmath0-spinor the spin and pseudospin are on equal footing , analogous to the two entangled @xmath119-spins in the bell state.@xcite since we have specified the forms of @xmath120 and @xmath121 in eq .
( [ eq : psi_basis ] ) and eq .
( [ eq : chi_basis ] ) , the coefficients in eq .
( [ eq : parametrizationz ] ) are complex numbers in general and the parameter @xmath109 is just the relative phase of these complex coefficients . throughout this paper
, we use the term `` entanglement '' as a synonym of the magnitude of the spin or pseudospin magnetizations : `` maximal entanglement '' means @xmath122 ; `` unentangled '' refers to @xmath115 or @xmath116 .
four types of qhfm ground states on the @xmath123 plane .
they are listed in tab .
[ tab : fm - tab ] . ] applying the parametrization in eq .
( [ eq : parametrizationz ] ) to the cp@xmath49-spinor @xmath86 representing the ferromagnetic state ( [ eq : qhfm ] ) , we minimize @xmath78 $ ] with respect to the six parameters of @xmath86 and obtain four types of qhfm ground states , which are displayed in fig .
[ fig : fm - region ] and tab .
[ tab : fm - tab ] . due to the zeeman coupling , a cp@xmath49-spinor @xmath86 with polarized spin along
the @xmath28-axis is always energetically favourable , independent of the pseudospin and entanglement .
it follows that the optimal value is @xmath124 in _ all regions _ throughout the @xmath123 plane , if we assume @xmath125 $ ] .
moreover , when @xmath126 ( region 1 , 3 ) , it is energetically favourable to orient the pseudospin along the @xmath28 ( or @xmath76 ) axis and one thus obtains an easy - axis fm state with optimal value @xmath127 or @xmath116 . on the contrary , when @xmath128 ( region 2 , 4 ) , a pseudospin magnetization in the @xmath31-plane is favoured , and one obtains an easy - plane fm state with optimal value @xmath129 and arbitrary optimal value for @xmath130 . on the line @xmath131 , the @xmath8 symmetry of pseudospin
is restored and the optimal values for @xmath132 and @xmath130 are arbitrary .
@xmath126 , @xmath133 unentangled easy - plane pspin fm & ( 2 ) @xmath128 , @xmath134 entangled fm with & ( 3 ) @xmath135preferential easy - axis pspin & entangled fm with & ( 4 ) @xmath136preferential easy - plane pspin & entanglement enriches the types of fm states because tuning the parameter @xmath114 allows the magnitudes of spin and pseudospin magnetization to decrease and thus to lower the anisotropic energy @xmath137 . when @xmath138 , the optimal value of @xmath114 is always 0 .
this is easily understood : since at least one of the parameters @xmath139 or @xmath83 is negative , @xmath140 is minimized when the spin and pseudospin magnetizations are maximal , i.e. when @xmath141 . for @xmath142 ,
the zeeman energy is crucial for the presence of unentangled fm states , which are of the same type as in the previous case .
were there no term proportional to @xmath143 in the expression of @xmath137 , all the other terms related to pseudospin would be non - negative , and the minimization would result in a spinor with @xmath144 ( maximal entanglement or @xmath122 ) for all possible @xmath145 in the 1-quadrant .
the term proportional to @xmath143 prefers larger magnitudes of @xmath146 along the applied magnetic field .
since @xmath147 always holds , there is a competition between the tendency towards small magnitude of pseudospin magnetization due to the pseudospin contribution to @xmath137 , and the tendency towards large magnitude of the spin magnetization due to the spin contribution . here in the case of @xmath142
, spin contributes more , so that a maximal value of @xmath148 is favoured , very much as for negative values of @xmath139 and @xmath83 , and spin and pseudospin remain unentangled .
the above - mentioned energy competition yields different results when @xmath149 . in this case
the pseudospin contribution to @xmath137 is positive , which can be lowered not only by proper choice of the direction ( @xmath113 ) of pseudospin magnetization , but also by shrinking the magnitude @xmath150 of pseudospin magnetization . although the later yield a less negative value of the spin contribution to @xmath137 due to the identity @xmath147 , the reduction of energy contribution from pseudospin overcomes the increase of the contribution from spin , so that the overall minimum of @xmath137 is reached when @xmath151 as a consequence of the identity between valley pseudospin and sublattice index in @xmath1 , the different qhfm bear a clear fingerprint in the spin - polarized electronic occupation of the two graphene sublattices .
the four types of qhfm ground states are visualized in figs .
[ fig : qhfm_uea ] , [ fig : qhfm_uep ] , [ fig : qhfm_eea ] and [ fig : qhfm_eep ] .
the upper parts ( panels a , b and c ) in each figure are the bloch sphere representations for the cp@xmath49-spinor @xmath86 of the qhfm ground state . in particular , panel ( a ) shows the spin magnetization @xmath152 ( eq . [ eq : si_y ] ) of the cp@xmath49-spinor @xmath86 in the spin bloch sphere . for the cp@xmath49-spinor @xmath86 corresponding to the unentangled qhfm states , with an easy - plane ( fig . [ fig : qhfm_uep ] ) and an easy - axis magnetization ( fig . [
fig : qhfm_uea ] ) , @xmath146 is a unit vector and its arrowhead is located _ on _ the spin bloch sphere .
in contrast , for the cp@xmath49-spinor @xmath86 corresponding to the entangled qhfm states , again with an easy - plane ( fig . [ fig : qhfm_eep ] ) and an easy - axis magnetization ( fig . [ fig : qhfm_eea ] ) , the vector @xmath146 has a magnitude @xmath153 smaller than @xmath117 , and explores the _ interior _ of the spin bloch sphere .
the appearance of the pseudospin magnetization @xmath154 in the pseudospin bloch sphere in panel ( b ) can be understood similarly . in the case of an easy - axis pseudospin ferromagnetic state that corresponds to the electronic occupation of a single sublattice , the arrow points along the @xmath28-axis , while a balanced occupation of both sublattices goes along with pseudospin magnetization pointing to the equator of the bloch sphere . again , in the presence of spin - pseudospin entanglement , the arrow starts to exploit the inside of the sphere .
panel ( c ) shows the _ entanglement vector _
@xmath155 for a cp@xmath49-spinor @xmath86 of the qhfm state , where @xmath114 and @xmath109 are obtained by the parametrization of @xmath86 with eq .
[ eq : parametrizationz ] .
the bounding sphere of the entanglement vector is called the _ entanglement bloch sphere _ by analogy . for the cp@xmath49-spinor @xmath86 corresponding to the unentangled qhfm states , both with easy - plane ( fig .
[ fig : qhfm_uep ] ) and easy - axis magnetization ( fig .
[ fig : qhfm_uea ] ) , the entanglement vector @xmath156 points to the north pole of the entanglement bloch sphere .
meanwhile , for the cp@xmath49-spinor @xmath86 corresponding to the entangled qhfm states ( figs .
[ fig : qhfm_eep ] and [ fig : qhfm_eea ] ) , the direction of @xmath156 is generic due to a non - vanishing value of @xmath114 . [ cols="^,^,^ " , ] visualization of the deflated cp@xmath49-pseudospin skyrmion with entangled easy - axis fm background ( point 12 in fig.[fig : skyrmion_type_diagram ] ) .
conventions are the same as in fig .
[ fig : pt4_spin_skyrmion ] . ]
we comment on the evidence of entanglement in a cp@xmath49-field @xmath157 .
although by examining the local patterns of the @xmath158 and @xmath159 profiles around a generic point @xmath160 in the @xmath31-plane , it is not possible to tell whether a cp@xmath49-field @xmath161 carries entanglement , it is possible to do so by examine those profiles at large scale over the entire texture . as we have discussed in detail , the factorizable skyrmions do not carry entanglement at any point in the texture , and the evidence is clear in the @xmath158 and @xmath159 profiles for a cp@xmath49-spin skyrmion , the pattern of sublattice occupation is uniform over the @xmath158 profile because of the constant pseudospin component , whereas for a cp@xmath49-pseudospin skyrmion , the two profiles are identical because of the constant spin component with polarization along the @xmath28-direction in the space of spin magnetization .
in contrast to the factorizable skyrmions , the entanglement skyrmions ( cp@xmath49-entanglement skyrmions , deflated cp@xmath49-pseudospin skyrmions ) embedded in the easy - axis pseudospin fm background usually have an `` anti - ferrimagnetic '' pattern in the @xmath159 profile , i.e. the spin magnetizations on two sublattices point in opposite directions . the cp@xmath49-entanglement skyrmion and the cp@xmath49-spin skyrmion which are embedded in the same easy - plane pseudospin fm background can not be distinguished by their @xmath158 and @xmath159 profiles .
but we can still make a difference by comparing their @xmath162 profiles .
in the previous section , we concentrated on a classification of the different skyrmion types one encounters in graphene qhfm in the @xmath1 ll .
the present section is devoted to a more quantitative analysis of the skyrmions size and energy , namely in the vicinity of phase transitions between the different fm background states .
size of optimal skyrmions at @xmath163 $ ] and @xmath164 .
black dashed lines highlight the boundaries of four regions for the four types of fm background of a skyrmion . ]
the size of a cp@xmath49 skyrmion is characterized by the optimal value of the @xmath165 parameter in the skyrmion ansatz . as discussed in sec .
[ subsec : skyrmion - ansatz ] , it is determined by the competition between the coulomb energy @xmath166 and the excess anisotropic energy @xmath167 . to be more precise , according to eq .
( [ eq : lambda_dep ] ) the optimal value of the @xmath165 parameter in the skyrmion ansatz is obtained by solving @xmath168 , which gives @xmath169}{2e_{\rm a}'[\tilde{z}_{\rm sk}]}\right)^{\frac{1}{3 } } , \label{eq : lambda_0}\ ] ] where @xmath170 is the @xmath0-field for an optimal skyrmion , but @xmath165 has been set to @xmath117 .
it has the same meaning as in eq .
( [ eq : lambda_dep ] ) in sec .
[ subsec : radial - deformation ] .
we define the skyrmion size @xmath171 ( in unit of @xmath44 ) by averaging @xmath172 over the topological charge density @xmath24 @xmath173 because @xmath24 can be viewed as the excess charge density induced by the texture in a cp@xmath49 skyrmion and is thus a good measure for the spatial extent of the texture . for a radially deformed skyrmion , we have @xmath174 with an @xmath172-dependent size parameter @xmath175 and topological charge density @xmath176^{2}}\left(1+\frac{2r^{2}}{\kappa\lambda_{0}^{2}}\right ) .
\label{eq : rho - topo - deformed}\ ] ] the coefficient @xmath177 is a monotonically increasing function of the radial deformation parameter @xmath178 .
its actual form depends on the radial deformation and is _ not _ related to the energy minimization . in the scale - invariant limit @xmath179 we have @xmath180 .
away from this limit , @xmath181 does not change much : for instance @xmath182 .
[ fig : skyrmion_size_all ] displays the size @xmath171 of optimal skyrmions obtained in sec .
[ subsec : minimization - results ] ( @xmath183 $ ] , @xmath164 ) .
one notices generally an increase in the skyrmion size for @xmath145 close to a transition between different ferromagnetic backgrounds .
these transitions are indicated by dotted black lines in fig .
[ fig : skyrmion_size_all ] . in order to investigate in more detail the increase in skyrmion size , we plot , in fig .
[ fig : skyrmsize ] , the skyrmion size @xmath171 as a function of @xmath184 along line a and line b. here , @xmath185 denotes the value at the border between two regions of different fm background , which is @xmath186 for line a separating the unentangled easy - axis from the unentangled easy - plane fm and @xmath187 for line b separating the unentangled from the entangled easy - axis fm .
one sees in fig .
[ fig : skyrmsize ] that along line a , the skyrmion size diverges ( red curves ) when approaching the transition at @xmath188 .
this is expected because , at the transition , the pseudospin texture costs the same amount of anisotropic energy as the state with uniform pseudospin magnetization , giving a vanishing excess anisotropic energy @xmath189 = e[\check{z}_{\rm sk } ] - e_{\rm a}[f]$ ] and thus a divergent @xmath165 .
physically this means that the pseudospin @xmath8 symmetry is restored when @xmath131 , and the coulomb energy of the skyrmion tends to inflate the skyrmion to infinite size . in the vicinity of the point @xmath190 , @xmath189 $ ] is proportional to @xmath191 .
consequently , @xmath189 \propto |u_{\rm z}-u_{\rm z0}|$ ] with @xmath192 , then from eq . ( [ eq : lambda_0 ] ) we know @xmath193
. therefore along line a , the skyrmion size @xmath171 scales as @xmath194 .
the exponent @xmath195 is extracted from the red lines of the log - log plot in the inset of fig .
[ fig : skyrmsize ] , in agreement with the above scaling argument .
the divergence of the skyrmion size at the line @xmath196 reflects the underlying transition between an easy - axis and an easy - plane pseudospin fm background , as long as both are unentangled .
skyrmion size ( in unit of magnetic length @xmath44 ) as a function of @xmath83 , along the lines a and b in fig .
[ fig : skyrmion_size_all ] . ] the blue curves in fig .
[ fig : skyrmsize ] show the evolution of the skyrmion size along line b cutting the border @xmath187 , where one also notices an enhanced albeit non - divergent size .
in contrast to the transition at @xmath188 , there is no symmetry restoration that would lead to a divergent skyrmion size , i.e. the anisotropy energy associated with the pseudospin remains finite here .
this yields a `` truncated power law '' , @xmath197 that is now cut off by a non - zero constant constant @xmath198 .
as one may expect from our simplified scaling analysis , the exponent @xmath199 is again close to @xmath200 .
even if there is no fully developed divergence in the skyrmion size , its increase unveils again a transition between different underlying fm background states .
the effect of disorder on skyrmions becomes significant when the length scale of the disorder potential is comparable to the skyrmion size . in such conditions ,
the charge - carrying skyrmion is trapped to the disorder potential that is coupled to electric charge .
moreover , if there is also pseudospin disorder in the hosting system , then a @xmath0-pseudospin skyrmion is frustrated by the two types of disorders . in the experiments where one can control the parameters @xmath82 and @xmath83 ,
the relation between the skyrmion size @xmath171 and the disorder length scale can be revealed in transport measurements .
energy of optimal skyrmions at @xmath163 $ ] and @xmath164 .
the dimensionless value @xmath201 is plotted in the figure , with the numerical value of @xmath202 .
black dashed lines highlight the boundaries of four regions for the four types of fm background of a skyrmion . ] as discussed in sec .
[ subsec : skyrmion - ansatz ] and sec . [ subsec : radial - deformation ]
, all scale - invariant @xmath0-skyrmions carrying topological charge @xmath203 have size - independent energy @xmath204 , and size - dependent energy @xmath205 .
the skyrmion size is determined by the competition between the coulomb energy @xmath166 and the anisotropic energy @xmath137 .
the radius deformation of skyrmion is introduced to make @xmath137 finite , at a cost of slight increase of @xmath206 and @xmath166 .
after minimization of the skyrmion energy at @xmath164 and @xmath183 $ ] , we find that the ratio @xmath207 shown in fig . [
fig : skyrmion_energy_all ] is between @xmath208 and @xmath209 , whereas the ratio @xmath210 is always close to @xmath117 .
thus the coulomb energy and anisotropic energy have significant contribution to the total energy of a @xmath0-skyrmion described by our model . despite of such contribution
, a @xmath0-skyrmion still has lower energy compared to the quasiparticle with a single spin / pseudospin flip on top of the qhfm state at quarter filling of the @xmath1 ll .
figure [ fig : skyrmion_energy_all ] shows the value @xmath211/4\pi\rho_s-1 $ ] , i.e. the relative energy difference between the optimal skyrmion @xmath212 , at different @xmath82 and @xmath83 , and the scale - invariant skyrmion , with energy @xmath213 .
we observe that the energy of an optimal skyrmion does not depend on @xmath82 and @xmath83 except when they are close to the lines @xmath214 or @xmath187 , as well as close to the line @xmath215 . as shown in the skyrmion phase diagram , fig .
[ fig : skyrmion_type_diagram ] , one obtains pseudospin skyrmions in the vicinity of the latter line .
their energy naturally depends on @xmath82 and @xmath216 in contrast to that of spin skyrmions that are encountered away from this line and that only depend on @xmath217 , which is constant here .
the energy of the spin skyrmions thus remains constant , as seen in the light orange parts of fig .
[ fig : skyrmion_energy_all ] that match the ( light red ) parts of the skyrmion phase diagram ( fig .
[ fig : skyrmion_type_diagram ] ) where one obtains pure spin skyrmions .
together with the plot of skyrmion size in fig .
[ fig : skyrmion_size_all ] , we also observed that , for different anisotropy parameter @xmath82 and @xmath83 , an _ optimal _
skyrmion has a lower _ optimal _ energy when its _ optimal _ size is larger .
the lowest value of @xmath211 $ ] is @xmath213 , which is achieved by the optimal skyrmion @xmath218 obtained at @xmath219 . at this condition
, the size of the optimal skyrmion diverges and its coulomb energy vanishes , because the skyrmion energy has a pseudospin @xmath8 symmetry .
the correlation between skyrmion energy and skyrmion size still holds when we vary the perpendicular magnetic field @xmath220 , which we choose here to be identical to the total magnetic field , i.e. we discard a tilt of field described in the following section .
this is shown in fig .
[ fig : energy_size_b ] , where we plot in a log - log scale the numerically obtained size and energy of four types of optimal skyrmions under @xmath221 .
different types of optimal skyrmion have different value of energy and size , but the shape of the lines are similar , which root from same scaling relations . indeed
eq . ( [ eq : lambda_0 ] ) implies that the skyrmion size ( in unit of @xmath44 ) scales as @xmath222 , and one obtains from eq .
( [ eq : delta_z - b ] ) then gives @xmath223 in natural units that are independent of the magnetic field .
one notices that the numerically obtained exponents in fig .
[ fig : energy_size_b](a ) agree to great accuracy ( within 2% ) with that obtained from our simple scaling analysis even if the latter does not take into account the radial deformation in eq .
( [ eq : lambda_r ] ) .
similarly , insert eq .
( [ eq : lambda_0 ] ) to eq .
( [ eq : lambda_dep ] ) , we obtain the scaling of the skyrmion energy @xmath224 or , in natural @xmath225-independent units @xmath226 which again agrees well ( within 2% ) with the numerically obtained scaling of the skyrmion energy plotted in fig .
[ fig : energy_size_b](b ) .
in the previous sections , we discussed the skyrmion type diagram , which provides a global view of the possible types of skyrmions at quarter filling of the @xmath1 ll of graphene monolayer . in this section
we show how this diagram is changed by modifications of the anisotropic energy .
we discuss three modifications to the anisotropic energy , each corresponding to one of the three contributions ( namely the in - plane pseudospin contribution , @xmath28-component pseudospin contribution and spin contribution ) .
the common way to modify the anisotropic energy is by sample tilting , where the sample is tilted away from the upright position while increasing the total magnetic field @xmath64 , so that the strength of the perpendicular component @xmath220 is kept to a constant value .
the electron spin is coupled to the total applied magnetic field , while the pseudospin contribution to the anisotropic energy is proportional to the component perpendicular to the sample .
therefore , sample tilting guarantees that the pseudospin contribution to the anisotropic energy is unaltered , but the electron spin contribution is increasing .
we further discuss terms that breaks the u@xmath227 symmetry between the @xmath228- and @xmath229-components of the pseudospin , whose contribution to the anisotropic energy is changed accordingly .
at last we discuss the pseudospin analogue of the zeeman coupling , i.e. the term proportional to @xmath230 , which describes the sublattice asymmetry and that could eventually be induced by substrates that are roughly commensurate with the carbon spacing in graphene ( e.g. boron - nitride substrates ) . the effect of sample tilting by an angle @xmath231 is the rescaling of the parameters @xmath139 , @xmath83 and @xmath217 in the anisotropic energy .
if we maintain our definition of the effective parameters @xmath232 in terms of the perpendicular component @xmath220 of the magnetic field , @xmath233 , and use the zeeman coupling , sensitive to the total magnetic field @xmath64 , as the overall energy scale , one obtains from eq .
( [ eq : enanisotropic ] ) @xmath234 & = & \frac{\delta_{\rm z}(b_{\rm t})}{2}\int \frac{d^{2}r}{2\pi l_{b}^{2 } } \times{\nonumber}\\ & & \left [ u_{\perp}\cos\varphi\left ( m_{\rm px}^{2}+m_{\rm py}^{2}\right)+u_{\rm z}\cos\varphi m_{\rm pz}^{2}-m_{\rm sz}\right],{\nonumber}\\\label{eq : enanisotropictilt}\end{aligned}\ ] ] where @xmath231 is the tilting angle from an upright direction , @xmath235 , and @xmath236 , @xmath237 are computed by eq .
( [ eq : ti_z ] ) , eq .
( [ eq : si_z ] ) respectively .
such a constant @xmath220 guarantees a constant length scale @xmath44 and a constant energy scale @xmath238 . after the rescaling of length and energy to unit @xmath44 and @xmath238 ,
@xmath239 is replaced by @xmath240 , where @xmath241 is computed by eq .
( [ eq : delta_z - b ] ) with @xmath242 , i.e. in the absence of sample tilting . at this stage , we see that the sample tilting by an angle @xmath231 changes the anisotropic energy as if we simply replace @xmath243 , and @xmath83 in the model without tilting by @xmath244 , and @xmath245 . to illustrate the effect of the tilt , we have plotted the evolution of the phase diagram in fig .
[ fig : skyrmion - type - tilting ] for a value of @xmath246 , as well as that for the skyrmion energy and size along line a in the insets .
one first notices that the tilt shifts the borders between the unentangled and entangled types of optimal fm states @xmath247 . generally the unentangled states are slightly favoured .
this can easily be understood as the consequence of the relative increase of the zeeman energy with respect to the pseudospin - symmetry - breaking terms the zeeman coupling favours indeed a full spin polarization and thus a maximal value of @xmath118 , which precisely means that the unentangled fm states are favoured .
change of the skyrmion type diagram for sample tilting by angle @xmath248 . shifted borders are marked by blue dashed lines for different types of fm backgrounds , and red for different types of skyrmions , in contrast to the grey and black lines which mark the borders in fig .
[ fig : skyrmion_type_diagram ] with @xmath249 .
the insets show the evolution of the skyrmion energy ( still in units of @xmath250 ) and size across line a. ] furthermore , the borders ( red lines in fig . [ fig : skyrmion - type - tilting ] for @xmath246 as compared to black lines for @xmath251 ) between the pseudospin skyrmions in the vicinity of @xmath252 are shifted .
indeed , the pseudospin skyrmions are favoured over a larger range as one may expect from the relative increase of the zeeman energy that renders larger the energy cost to create a spin skyrmion .
this is clearly seen in the inset of fig .
[ fig : skyrmion - type - tilting ] while the energy of the pseudospin skyrmion , in the tilt - independent units of @xmath250 , remains the same as for @xmath253 , the @xmath254-independent energy of the spin skyrmion is increased and its size reduced such that the region of stable pseudospin skyrmions is increased .
notice finally that the scaling relations ( [ scalingrb ] ) and ( [ scalingeb ] ) are also affected by the sample tilting .
the scaling of the size of pseudospin skyrmions still obeys the @xmath255 law because the pseudospin anisotropy depends only on @xmath62 .
we can also express such scaling in the total magnetic field and the tilting angle , namely @xmath256 by contrast , the spin skyrmion follows a scaling relation @xmath257 the exponents on @xmath258 and the magnetic field @xmath62 or @xmath64 are different because the coulomb energy @xmath166 depends on @xmath259 while the zeeman coupling depends on @xmath260 .
if one perform the sample tilting at constant total magnetic field @xmath64 , spin and pseudospin skyrmions would show a different scaling with respect to the tilt angle . change of the skyrmion type diagram for @xmath261 .
the color code is the same as in fig .
[ fig : skyrmion - type - tilting ] , in the absence of tilt ( @xmath253 ) . ] our model for the anisotropic energy @xmath137 in eq .
( [ eq : enanisotropic ] ) [ also eq . (
[ eq : enanisotropictilt ] ) ] has a u@xmath227 symmetry for the @xmath228- and @xmath229-component of the pseudospin .
formally , this symmetry can be broken with the help of the parameter @xmath262 , which yields the modified energy @xmath234 & = & \frac{\delta_{\rm z}(b_{\rm t})}{2}\int\frac{d^{2}r}{2\pi l_{b}^{2}}\left[(u_{\perp } + \delta u_{\perp})\cos\varphi m_{\rm px}^{2}\right .
{ \nonumber}\\ & & \left .
+ ( u_{\perp}-\delta u_{\perp})\cos\varphi m_{\rm py}^{2 } + u_{\rm z}\cos\varphi m_{\rm pz}^{2}-m_{\rm sz}\right],{\nonumber}\\\label{eq : enanisotropicsubstrateaniso}\end{aligned}\ ] ] where we have also considered sample tilting . it is tempting to attribute this symmetry breaking to an @xmath31-anisotropy in the graphene sheet , e.g. when one applies uni - axial strain to it that breaks the @xmath263 rotation symmetry of the lattice .
however , more microscopic calculations beyond the scope of our paper , would be required to corroborate this picture .
the main effect of this symmetry breaking is to favour , in the case of an easy - plane pseudospin fm , an orientation of the pseudospin magnetization along the @xmath228-axis ( @xmath264 ) for @xmath265 or along the @xmath229-axis ( @xmath266 ) when @xmath267 , while no particular direction is favoured in the isotropic case ( @xmath268 ) .
the relevant energy scale for the in - plane anisotropy is therefore reduced , and we can use the above model ( [ eq : enanisotropictilt ] ) also to treat the present case if one replaces the model parameters @xmath145 by @xmath269 . the evolution of the phase diagram is shown in fig . [
fig : skyrmion - type - substrate - anisotropy ] for a value @xmath270 .
one first notices that the transition lines towards the easy - plane fm states is modified , whereas the easy - axis fm states are naturally not affected in energy by the change in @xmath82 .
indeed , the transition line between the unentangled easy - plane and the easy - axis fm states is shifted downwards the transition occurs at @xmath271 instead of @xmath188 as in the isotropic case .
furthermore , the zeeman energy is again effectively increased with respect to the easy - plane fm energy scale ( for positive values of @xmath139 ) such that the unentangled easy - plane fm states are stabilized , whence the right shift of the transition between the unentangled and entangled easy - plane fm states in fig .
[ fig : skyrmion - type - substrate - anisotropy ] .
similarly one may understand the borders of the regions for the different skyrmion types . while the transition between the spin and the pseudospin skyrmion is unaffected in the case of an easy - axis fm background characterized by the unaltered energy scale @xmath83 , that between the spin an pseudospin skyrmions in an easy - plane fm background is shifted downwards ( red line in fig .
[ fig : skyrmion - type - substrate - anisotropy ] . this can be understood from the evolution of the skyrmion size shown in the lower inset ( cut along line a ) .
when approaching the transition between the two different underlying fm states from the right hand side ( by lowering the value of @xmath139 , the skyrmion size follows first the same evolution as in the isotropic case until the maximum is reached at the value of the underlying transition .
notice , however , that due to the in - plane anisotropy the pseudospin @xmath8 symmetry is no longer restored so that the skyrmion size does not diverge . left to the transition , the skyrmion size is therefore generically smaller than in the isotropic case , and the skyrmion energy higher ( see upper inset of fig . [
fig : skyrmion - type - substrate - anisotropy ] ) .
the energy of the spin skyrmion is therefore reached at smaller values of @xmath272 such that the region where one encounters pseudospin skyrmions becomes smaller .
furthermore , the transition between the region where one finds entanglement skyrmions and that of deflated pseudospin skyrmions in the region @xmath273 is strongly affected by the in - plane anisotropy .
this can be attributed to the very small energy difference and thus delicate competition between entanglement and deflated pseudospin skyrmions that are also characterized by a certain degree of entanglement , as discussed in sec .
[ subsec : deflated - pseudospin - skyrmion ] .
change of the skyrmion type diagram for @xmath274 .
the color code is the same as in fig .
[ fig : skyrmion - type - tilting ] , in the absence of tilt ( @xmath253 ) . ] the anisotropic energy eq .
( [ eq : enanisotropic ] ) can also be modified by adding the term @xmath275 , which is a pseudospin analogue of the zeeman coupling and describes the sublattice asymmetry . omitting the @xmath262 term discussed in the previous section but including the sample tilting , we have @xmath234 & = & \frac{\delta_{\rm z}(b_{\rm t})}{2}\int\frac{d^{2}r}{2\pi l_{b}^{2 } } \left [ u_{\perp}\cos\varphi\left ( m_{\rm px}^{2}+m_{\rm py}^{2}\right)\right .
{ \nonumber}\\ & & \left.+ u_{\rm z}\cos\varphi m_{\rm pz}^{2 } + \delta u_{\rm z}\cos\varphi m_{\rm pz } -m_{\rm sz}\right],{\nonumber}\\\label{eq : enanisotropicpspinzeeman}\end{aligned}\ ] ] where @xmath276 $ ] is the relative energy scale of the pseudospin analogue of the zeeman coupling . the original @xmath75 symmetry ( equivalence of a pseudospin orientation along the @xmath28 or @xmath76-axis ) is now broken , and pseudospin orientation along @xmath28 are favoured for @xmath277 and along @xmath76 for @xmath278 .
the @xmath279 term modifies the optimal fm states @xmath247 for the anisotropic energy in a complementary manner as @xmath262 instead of the easy - plane fm states , the easy - axis fm states are affected by this type of anisotropy .
this can indeed be seen in the phase diagram shown in fig .
[ fig : skyrmion - type - pseudospin - zeeman ] for @xmath274 .
first , it shifts the border between the regions for unentangled easy - axis fm and for unentangled easy - plane fm , as well as the border between the regions for unentangled easy - axis fm and entangled easy - axis fm .
both are shifted upwards by @xmath280 .
second , the slope of the border between the regions for entangled easy - axis fm and for entangled easy - plane fm , is increased from @xmath117 to @xmath281 .
moreover , at the upper - left side of the sloped borders separating the easy - axis fm and easy - plane fm regions , the @xmath282 points close to the border corresponds to fm spinor @xmath86 with optimal value of @xmath132 less than @xmath283 . as @xmath282 moves away from these borders ( to the upper - left side ) , @xmath132 approaches to @xmath283 .
these modifications can be understood by rewriting @xmath284 $ ] as @xmath234 & = & \frac{\delta_{\rm z}(b_{\rm t})}{2}\int\frac{d^{2}r}{2\pi l_{b}^{2}}\times\nonumber\\ & & \left[u_{\perp}\cos\varphi\left(m_{{\rm px}}^{2}+m_{{\rm py}}^{2}\right)+\left(u_{{\rm z}}+\frac{\delta u_{{\rm z}}}{2}\right)\cos\varphi m_{{\rm pz}}^{2}\right.\nonumber\\ & & \left.-\frac{\delta u_{{\rm z}}}{2}\cos\varphi(m_{{\rm pz}}-1)^{2}-m_{{\rm sz}}+\frac{\delta u_{{\rm z}}}{2}\cos\varphi\right ] \label{eq : enanisotropicpspinzeeman1}\end{aligned}\ ] ] the analysis is similar to those for the the @xmath262 term for the substrate anisotropy in the previous subsection .
the upward shift of the transition between the easy - axis and easy - plane fm states is a consequence of the lowered energy of the former ( for @xmath285 ) that favours easy - axis states .
similarly to the easy - plane anisotropy discussed above , there is no restoration of the @xmath8 pseudospin symmetry at the transition line @xmath286 because of the remaining term @xmath287 in eq .
( [ eq : enanisotropicpspinzeeman1 ] . as a consequence ,
the skyrmion size does not diverge at this value and its energy does not reach the minimal ( scale - invariant ) value @xmath288 ( see insets of fig .
[ fig : skyrmion - type - pseudospin - zeeman ] ) .
in conclusion , our analysis of the quarter - filled @xmath1 ll of graphene shows that the anisotropic terms in the energy functional , which break the spin and valley pseudospin symmetry of the leading coulomb energy , lead to four types of ferromagnetic ground state and various types of @xmath0-skyrmions .
a skyrmion is embedded in the ferromagnetic background @xmath86 , and the center spinor @xmath198 gives rise to different types of skyrmions because the orthogonality condition between the spinors , @xmath289 , does not entirely fix the skyrmion type .
the favored skyrmion type is then obtained by a minimization of the anisotropic energy terms .
when the electron spin and pseudospin contribution to the anisotropic energy varies , a spin skyrmion can change to a pseudospin skyrmion while the ferromagnetic background remains unaltered .
this is strikingly different from the two - fold degenerate ll split by electron spin , where a skyrmion simply disappears when the magnetic field , and hence the zeeman coupling increases to a critical value .
a large spin skyrmion requires weak zeeman coupling and hence a weak applied magnetic field .
however , in our four - component system where @xmath0-skyrmions are formed , one can tune the model parameters @xmath290 to approximately restore the @xmath8 symmetry of the valley pseudospin , so as to have a large pseudospin skyrmion , even when the anisotropic energy for both spin and valley pseudospin is high . by analogy
, skyrmion also appears in graphene monolayer at @xmath291 when the valley pseudospin @xmath8 symmetry or the @xmath292 symmetry@xcite is approximately restored .
in addition to these skyrmion types , one encounters more exotic skyrmions that are formed due to spin - pseudospin entanglement .
a direct transposition of the concept of a skyrmion that entirely covers the spin or pseudospin bloch sphere yields the _ entanglement _ skyrmion , which covers the third _ entanglement _ bloch sphere .
this bloch sphere is required , at @xmath293 , to fully characterize the skyrmion and fm states i.e. to account for two different angles @xmath114 and @xmath109 that naturally arise in our @xmath0 description of @xmath26 states . in the case of easy - axis fm states , entanglement manifests itself , rather counter - intuitively , in the form of locally anti - ferromagnetic or anti - ferrimagnetic patterns .
moreover , we have identified another type of skyrmion , a deflated pseudospin skyrmions with partial entanglement .
while all directions in the pseudospin bloch sphere are explored within this skyrmion type , similarly to an unentangled pseudospin skyrmion , the magnitude of the pseudospin magnetization is reduced in most places .
this provides the deflated image of the representation on the bloch sphere , accompanied by an exploration of parts of the spin bloch sphere . in lattice - resolved images ,
deflated pseudospin skyrmions are manifest by a lower spin - pseudospin contrast and by local deviations of the spin magnetization from the @xmath28-axis .
the lattice - resolved images we obtain in our study can in principle be used as a guide in a possible experimental identification of the various skyrmions in graphene , e.g. in spin - resolved stm / sts measurements . while the varying spin magnetization can be probed by a combination of magnetic tips , not only in the @xmath28-direction but also in the @xmath228- and @xmath229-directions ,
the @xmath28-component of the pseudospin magnetization can directly be read of from the relative electronic occupation of the different sublattices in graphene .
in contrast to the spin magnetization , the @xmath228- and @xmath229-components of the pseudospin magnetization are unfortunately not accessible in this type of experiments that do not provide insight into the combination @xmath294 of the parameters in our parametrization of @xmath0 fields .
our analysis also shows that the model parameters @xmath82 and @xmath83 that correspond to large optimal @xmath0-skyrmions reside in the vicinity of the borders between two different types of ferromagnetic backgrounds .
indeed , the transitions are characterized by either ( i ) a partial symmetry restoration , e.g. of the @xmath8 pseudospin symmetry at @xmath295 , or ( ii ) a tendency of such a symmetry restoration , e.g. at @xmath187 for @xmath296 or @xmath297 for @xmath298 between unentangled and entangle fm backgrounds . in general
, the energy to create a skyrmion is reduced due to a reduction of the anisotropy energy . in the case ( i )
, the partial symmetry restoration yields a divergent skyrmion size described by a power law @xmath299 , with a critical exponent @xmath199 close to @xmath200 , as expected from simple scaling arguments . at the same time , the skyrmion size in @xmath225-independent units scales in the vicinity of the transition as @xmath300 with the @xmath225-field and its energy as @xmath301 when compared to the energy @xmath302 of the non - linear sigma model .
right at the transition , the skyrmion energy retrieves its scale - invariant value @xmath302 . if there is only a tendency towards symmetry restoration [ case ( ii ) ] , either in the above mentioned transition between unentangled and entangled fm backgrounds or in the presence of an anisotropy in the substrate that couples to the pseudospin , the divergence in the skyrmion size is truncated and follows approximately the law @xmath303 , where @xmath198 is some positive constant and @xmath304 is a generic coupling constant with critical value @xmath305 .
again we find that the exponent @xmath199 is close to @xmath200 .
several terms are not included in our model .
the @xmath26 invariance of the non - linear sigma model energy @xmath306 $ ] can be broken by a term of the form@xcite @xmath307 $ ] , where @xmath308 is a superposition of the tensor product of pauli matrices , and the coefficient @xmath309 is of the same order as the input parameters @xmath82 , @xmath83 for the anisotropic energy in our model .
at the fourth order of the gradient expansion , which may be relevant for small skyrmion sizes , we may have@xcite @xmath310 ^ 2 $ ] , where @xmath311 is the laplacian .
the minus sign of such a term is on the same order as the coulomb energy @xmath312 $ ] in the gradient expansion , and it implies that our model has overestimated@xcite the energy of a skyrmion , or equivalently , underestimated the skyrmion size
. generally speaking , the non - linear sigma model of skyrmions does not take account for the reduction of magnetization in the core of the texture caused by quantum fluctuations.@xcite
we most gratefully acknowledge fruitful discussions and former collaboration with achim rosch .
furthermore , we thank markus morgenstern for helpful experimental insights .
yl is funded by a scholarship from the china scholarship council .
in sec . [ sec:4-types - skyrmion ] we represent the @xmath0 textures @xmath313 in lattice - scale profiles of the electron density @xmath158 and the @xmath28-component of the spin magnetization @xmath159 for the @xmath0-field @xmath313 .
these profiles are computed with the help of eqs .
( [ eq : si_z ] ) and ( [ eq : ti_z ] ) , where the components of the @xmath0-field are convoluted with a form factor @xmath314 that mimics the amplitude square of the atomic wave function . here
, @xmath315 is the position on the sublattice @xmath316 in the @xmath317-th unit cell . for illustration purposes
, we have chosen a gaussian profile for @xmath318 that is normalized as @xmath319 . moreover ,
if we neglect the overlap between the atomic wave functions centred around different lattice site in the function @xmath320 , the expressions for @xmath158 and @xmath159 simply read @xmath321 the electronic occupation of the sublattices a and b in the @xmath317-th unit cell is given by @xmath322 , while the @xmath28-component of the spin magnetization is @xmath323 , with @xmath324z({{\bf r}}_{j}^{a , b}),\\ s_{\rm z}({{\bf r}}_{j}^{a , b } ) & = z^{\dagger}({{\bf r}}_{j}^{a , b})\left[\frac{\sigma_{z}\pm1}{2}\otimes\sigma_{z}\right]z({{\bf r}}_{j}^{a , b}).\end{aligned}\ ] ] in the above expression , the @xmath325-sign in the projector is chosen for a site on the @xmath326-sublattice and @xmath327 for a site on the @xmath225-sublattice .
the second homotopy group@xcite @xmath329 of a topological manifold @xmath330 is defined as the equivalence class of the continuous mapping @xmath331 , where two mappings @xmath332 and @xmath333 are equivalent means that there exists a continuous mapping @xmath334\times s^2\rightarrow { \cal m}$ ] such that @xmath335 and @xmath336 .
the multiplication in the group is the homotopic composition of the continuous mappings .
the second homotopy group @xmath12 of the 2-sphere @xmath11 can be understood as `` wrapping a 2-sphere on another 2-sphere '' . the winding number ( brouwer degree ) of the mapping @xmath337 classifies all possible ways of the `` wrapping '' . since @xmath338 , we have @xmath339 .
it can also be computed@xcite by the following relations between a simply connected lie group @xmath340 , one of its subgroups @xmath341 and the coset space @xmath342 : @xmath343 where @xmath344 is the fundamental group of a manifold defined by replacing @xmath11 in the definition of @xmath329 by @xmath13 .
the manifold @xmath345 is isomorphic to @xmath346 , therefore @xmath347 where the last equation @xmath348 follows from the classification of mappings from a closed path to another closed path .
the manifold @xmath328 is isomorphic to @xmath349 \cong \su{4}/[\su{3}\times { \rm u}(1)],\ ] ] therefore @xmath350 ) = \pi_1(\su{3}\times { \rm u}(1)).{\nonumber}\ ] ] using the fact that the homotopy group for product manifold factorizes , i.e. @xmath351 and the fact that any simple lie group @xmath340 has @xmath352 , we obtain @xmath353 this result can be generalized@xcite to compute @xmath354 for integer @xmath355 , because @xmath356 ) { \nonumber}\\ & = & \pi_1(\su{{\mathcal{n}}-1}\times { \rm u}(1)){\nonumber}\\ & = & \pi_1(\su{{\mathcal{n}}-1})\times \pi_1({\rm u}(1)){\nonumber}\\ & = & \mathbb{z}\end{aligned}\ ] ] this argument shows that any @xmath16-component qh system or , more generally , any @xmath16-component ferromagnet in two spatial dimensions with one filled component can principally host topological textures in the form of skyrmions that are characterized by an integer topological charge @xmath357 .
the specificity of the quantum hall ferromagnet is that this topological charge is directly related to an electric charge .
the @xmath0 space is a collection of normalized , four - component complex vectors @xmath358 . in addition , two vectors @xmath359 and @xmath360 are equivalent for arbitrary @xmath361 , in the sense that they correspond to the same matrix @xmath362 , which plays the role of order parameter in the theory of qhfm in ref . [ ] . to be more precise , @xmath359 should be understood as a representative of the equivalent class @xmath363\in v/\sim$ ] , where @xmath364 denotes the set of normalized @xmath365 vectors and @xmath366 denotes the equivalence relation @xmath367 for @xmath368 and @xmath361 . in this paper , we call the normalized @xmath365 vector @xmath359 a `` @xmath0-spinor '' if it represents an element in the @xmath0 space .
the name `` spinor '' here is inherited from the physics literature without mathematical rigour .
the @xmath0 space can also be understood as a coset space ( see eq . ( [ eq : group ] ) in app .
[ subsec : second - homotopy ] ) @xmath369 \cong \su{4}/[\su{3}\times { \rm u}(1)],\ ] ] which means that the integer filling of one of the four landau sublevels is distinguished by removing the @xmath370 ( or @xmath371 ) rotations among three empty sublevels from the entire rotation group @xmath372 ( or @xmath26 ) among all four sublevels , followed by a removal of a global phase factor @xmath373 .
the parametrization of a cp@xmath49-spinor is discussed in sec .
[ subsec : qhfm - ground - states ] .
there we construct the cp@xmath49-spinor from six angles @xmath374 , @xmath375 , @xmath132 , @xmath130 , @xmath114 , @xmath109 , and visualized them via a triplet of bloch spheres .
the inverse problem to obtain the six angles parametrizing a given cp@xmath49-spinor up to an overall phase factor @xmath376 is equally important for the complete understanding of the parametrization eq .
( [ eq : parametrizationz ] ) . given a cp@xmath49 -spinor @xmath96 , we have access to the parameter @xmath114 by computing the magnitude of spin and pseudospin magnetization via eqs .
( [ eq : si_f ] ) and ( [ eq : ti_f ] ) .
there are two possible values : @xmath377 and @xmath378 .
they correspond to the same cp@xmath49-spinor because of the following equivalence between two cp@xmath49-spinors parametrized in eq .
( [ eq : parametrizationz ] ) : @xmath379 the effect of changing @xmath114 into @xmath380 while keeping the same cp@xmath49-spinor is to reverse the direction of the spin and pseudospin magnetization .
it reveals the fact that our parametrization is redundant for @xmath381 . in the discussion of qhfm states
, we can always restrict @xmath114 in the range @xmath382 $ ] . under this restriction , we have @xmath383 and @xmath384 , in agreement with the schmidt decomposition theorem , which claims that we have non - negative real numbers as the coefficients in front of the direct - product - state basis obtained from the decomposition .
notice , however , that this restriction is not the unique choice to avoid redundancies in the parametrization . instead one can also choose the restriction @xmath374 or @xmath132 in @xmath382 $ ] , such that @xmath114 and @xmath109 span the full bloch sphere , as in the discussion of entanglement skyrmions in sec .
[ subsec : entanglement - skyrmion ] .
the cp@xmath49-non - linear sigma model ( cp@xmath49-nlsm ) has the following form of energy : @xmath385=\int d^2r { { \bf d}}z^{\dagger}({{\bf r}})\cdot { { \bf d}}z({{\bf r}}),\label{eq : nlsm - general}\ ] ] where we defined @xmath386 z({{\bf r}})$ ] .
the finite - energy configurations of @xmath157 satisfies the boundary condition at spatial infinity : @xmath387 where @xmath388 is a function of the polar angle @xmath231 of coordinate plane , and @xmath389 is a constant cp@xmath49-spinor . under such boundary condition
, the base manifold of field @xmath157 can be extended to the riemann sphere @xmath390 via riemann stereographic projection,@xcite and the value of the cp@xmath49-spinor field at spatial infinity is @xmath391 .
after the extension , each field @xmath157 can be considered as a mapping @xmath392 and can be classified by the degree of mapping @xmath393 & = \int\left(-\frac{i}{2\pi}\right)[{{\bf d}}z({{\bf r}})^{\dagger } \times { { \bf d}}z({{\bf r}})]_z d^{2}r\nonumber\\ & = \int\rho_{\rm topo}({{\bf r}})d^{2}r\end{aligned}\ ] ] because the second homotopy group @xmath394 of the cp@xmath49 space ( as target space of field @xmath157 ) is non - trivial .
@xmath7 is called _ topological charge _ in the physics literature , and correspondingly , @xmath24 is called _ topological charge density_. it is impossible to continuously deform a field @xmath395 into another configuration @xmath396 of different topological charge @xmath397\neq{\cal q}[z_{a}]$ ] . therefore , our variational analysis on @xmath398 $ ] ( eq .
( [ eq : nlsm - general ] ) ) is limited within a subspace of field configurations @xmath399=m\in\mathbb{z}\right\ } .\ ] ]
the minimal - energy configuration @xmath400}({{\bf r}})$ ] for @xmath398 $ ] in each @xmath401 serves as the starting point of our analysis of cp@xmath49-skyrmions .
we observe an inequality @xmath402\pm4\pi{\cal
q}[z]\ge 0\end{aligned}\ ] ] or , equivalently , @xmath385\ge2\pi\left|{\cal q}[z]\right|.\ ] ] it means that for each @xmath401 , the lower bound of @xmath403 $ ] is @xmath404\right|$ ] . comparing eq .
( [ eq : nlsm - general ] ) to eq .
( [ eq : ennlsm ] ) to restore the units of energy , we find that the right hand side in the above inequality gives @xmath405 as the lower bound of @xmath206 for charge @xmath406 skyrmions . moreover , the minimal - energy configuration which saturates the inequality is the solution of the following first order partial differential equations , which are called the bogomolnyi - prasad - sommerfield ( bps ) equations : @xmath407<0,\label{eq : bps_minus}\\ d_{i}z - i\epsilon_{ik}d_{k}z & = 0\;\mbox { for } { \cal q}[z]>0.\label{eq : bps_plus}\end{aligned}\ ] ] inserting @xmath408 with the _ unormalized _ complex vector @xmath409 and using the complex coordinate @xmath410 , the above equations are rewritten as @xmath411 where @xmath412 and @xmath413 . the topological charge @xmath7 can also be expressed in @xmath414 with the help of stoke s theorem : @xmath415 where the contour @xmath416 encloses a simply connected region @xmath417 , and @xmath417 maximally covers the punctured complex plane where @xmath414 and @xmath418 are well defined .
the generic solution of bps equation eq .
( [ eq : bps_plus ] ) for @xmath419 ( @xmath420 ) is a four - component vector field @xmath421 , in which the components @xmath422 are meromorphic ( anti - meromorphic ) functions . in the main text
we are interested in the solution of bps equation with topological charge @xmath203 .
the generic form is @xmath423}(z ) = \left [ f_1 z - c_1 , f_2 z - c_2 , f_3 z - c_3 , f_4 z - c_4 \right].\ ] ] however , to demonstrate the use of the above formula in eq .
( [ eq : topou1charge ] ) , let us compute @xmath7 for the following complex vector field @xmath424,\label{eq : w - example}\ ] ] where the topological charge can not be simply read off .
the contour @xmath425 consists three parts an anticlockwise circle @xmath426 , a clockwise circle @xmath427 , and a pair of straight lines @xmath428 of opposite directions connecting the two circles .
we have @xmath429 \frac{w^{\dagger}\boldsymbol{\nabla}w}{w^{\dagger}w}\cdot d\boldsymbol{l } { \nonumber}\\ & = & \frac{1}{2\pi i}\left [ \left(0 + 0 + \frac{2\pi i}{2 } + \frac{2\pi i}{2}\right ) - \left(0 - 2\pi i + 0 + 0\right ) + \left(0\right ) \right]{\nonumber}\\ & = & 2,{\nonumber}\end{aligned}\ ] ] where each term in the round bracket corresponds to the contribution from the components of @xmath430 in eq .
( [ eq : w - example ] ) accordingly .
it is easy to verify the above result for the topological charge of @xmath430 in eq .
( [ eq : w - example ] ) by directly computation with eq .
( [ eq : topologicaldensity ] ) in the main text .
in sec . [ subsec : radial - deformation ] we discussed the radial deformed skyrmion @xmath212 .
it has the form @xmath431}/\sqrt{\check{w}_{\rm [ 1]}^{\dagger}\check{w}_{\rm [ 1]}}\ ] ] with @xmath432}(x , y ) = ( x+iy)f-\lambda(r)c$ ] and @xmath433 . in the limit @xmath434 , the three components of the energy @xmath435 $ ]
have the following behavior @xmath436 & \rightarrow & 4\pi\sqrt{\frac{\pi}{32 } } \frac{e^{2}}{4\pi\epsilon l_{b}},\\ e_{\rm c}[\check{z}_{\rm sk } ] & \rightarrow & \frac{3\pi}{64 } \frac{e^{2}}{4\pi\epsilon l_{b } } \left(\frac{\lambda_0}{l_b}\right)^{-1},\\ e_{\rm a}[\check{z}_{\rm sk } ] & \sim & \left ( a_{0}+a_{1}\left|\delta_{\rm z}\right|\log\kappa\right)\left(\frac{\lambda_0}{l_b}\right)^{-2}.\end{aligned}\ ] ] as a function of @xmath178 , @xmath437 and @xmath166 decrease monotonically .
meanwhile , @xmath137 is a monotonic increase function of @xmath178 .
[ fig : kappa_dependence ] displays their @xmath178 dependence .
based on this analysis , the radial - deformation ansatz @xmath212 gives a finite anisotropic energy @xmath137 and a lowered total energy .
@xmath178-dependence of the nlsm energy @xmath206 ( red line ) , the coulomb interaction energy @xmath166 ( blue line ) and the anisotropic energy @xmath137 ( green lines ) . ]
the @xmath345-skyrmion ( or o(3)-skyrmion ) can be visualized by the image of the riemann sphere on the bloch sphere .
the riemann sphere is obtained by the stereographic projection between the cartesian coordinates @xmath438 on the plane and the cartesian coordinates @xmath439 on the riemann sphere : @xmath440 it is illustrated in fig .
[ fig : riemann_sphere ] .
the `` hedgehog '' skyrmion is in fact an identity mapping from the riemann sphere to the bloch sphere . as discussed in sec .
[ subsec : qhfm - ground - states ] and appendix .
[ subsec : cp3-spinor - and - nlsm ] , a cp@xmath49-spinor can be parametrized by six angles @xmath374 , @xmath375 , @xmath132 , @xmath130 , @xmath114 , @xmath109 , and therefore can be visualized by three bloch spheres , namely the spin bloch sphere ( for @xmath374 and @xmath375 ) , the pseudospin bloch sphere ( for @xmath132 and @xmath130 ) and the entanglement bloch sphere ( encodes @xmath114 and @xmath109 ) .
hence a cp@xmath49-skyrmion can be visualized by plotting the image of the riemann sphere on the three bloch spheres via the parametrization of the cp@xmath49-spinor . in practice , we draw the spin magnetization and pseudospin magnetization @xmath441 at points specified by longitudinal and latitudinal lines on the riemann sphere .
these lines correspond to rays of constant angle @xmath48 and circles of constant radius @xmath172 circles on the @xmath31-plane where the skyrmion lives .
in general , at a given point @xmath442 on the @xmath31-plane , @xmath443 and @xmath444 may have a magnitude less than @xmath117 , because @xmath445 can be different from @xmath446 or @xmath116 . as a consequence on the visualization , the spin and pseudospin magnetization lay _ inside _ the corresponding bloch spheres .
this is the case for the cp@xmath49-entanglement skyrmion and the deflated cp@xmath49-pseudospin skyrmion .
for example , in chapter 10 of l. h. ryder , _ quantum field theory _
( second edition ) , cambridge university press 1996 ; chapter 6 of n. manton and p. sutcliffe , _ topological solitons _ , cambridge university press 2004 .
s. mhlbauer , b. binz , f. jonietz , c. pfleiderer , a. rosch , a. neubauer , r. georgii , and p. bni , . n. romming , c. hanneken , m. menzel , j. e. bickel , b. wolter , k. von bergmann , a. kubetzka , and r. wiesendanger , . s. m. girvin and a. h. macdonald , chapter 5 in _ perspectives in quantum hall effects : novel quantum liquids in low - dimensional semiconductor structures _ , editor(s ) : sankar das sarma , aron pinczuk wiley - vch verlag gmbh & co. kgaa ( 2004 ) . | we model quantum hall skyrmions in graphene monolayer at quarter filling by a theory of @xmath0 fields and study the energy minimizing skyrmions in presence of valley pseudospin anisotropy and zeeman coupling .
we present a diagram of all types of skyrmions in a wide range of the anisotropy parameters . for each type of skyrmion
, we visualize it on three bloch spheres , and present the profiles of its texture on the graphene honeycomb lattice , thus providing references for the stm / sts imaging of spin - pseudospin textures in graphene monolayer in quantum hall regime .
besides the spin and pseudospin skyrmions for the corresponding degrees of freedom of an electron in the @xmath1 landau level , we discuss two unusual types the `` entanglement skyrmion '' whose texture lies in the space of the entanglement of spin and pseudospin , as well as the `` deflated pseudospin skyrmion '' with partial entanglement . for all skyrmion types ,
we study the dependence of the energy and the size of a skyrmion on the anisotropy parameters and perpendicular magnetic field . we also propose three ways to modify the anisotropy energy , namely the sample tilting , the substrate anisotropy and the valley pseudospin analogue of zeeman coupling . |
xte j0421 + 560 was discovered by the all - sky monitor ( asm ) on the rossi x - ray timing explorer ( rxte ) on 31.64 march 1998 @xcite as the first bright x - ray transient in several years , reaching almost 2 crabs within a day of the first detection .
corresponding radio emission was detected soon after ( 1.9 april ; @xcite ) .
optical observations @xcite identified the radio and x - ray transient as the star , based on positional coincidence with the radio emission , and a concurrent optical outburst .
it was also detected by the burst and transient source experiment in the low energy gamma
rays , reaching a flux of 1.6mcrab in the 2040 kev band @xcite . is a peculiar bright emission - line star with unusually strong fe ii lines @xcite , indicating a very hot star or stellar shell .
it also has a near - ir excess @xcite . has a rich and unusual optical spectrum and
has been identified as a symbiotic star @xcite , a herbig be star @xcite , a b[e ] star @xcite , and a supergiant b[e ] ( sgb[e ] ) star @xcite .
the recent consensus seems to be that it is either a b[e ] or sgb[e ] star .
if it is an sgb[e ] star it would be the only one identified in our galaxy , all others having been found in the magellanic clouds .
its distance has been estimated to be anything from 1kpc @xcite to 8kpc ( e.g. , @xcite ) based on luminosity , spectrum , interstellar absorption , and kinematics .
here we take 5kpc as a reasonable compromise . was not previously a known radio source , with two 10.7ghz observations in 1973 giving upper limits of 5mjy @xcite .
s x - ray / radio / optical flare puts it into another class , that of x - ray novae .
if is at a distance greater than 2kpc then its x - ray luminosity indicates that there is probably a compact object ( black hole or neutron star ) in the system ( e.g. , @xcite ) .
@xcite suggests that the outburst was the result of thermonuclear runaway on the surface of a white dwarf , which would require a closer distance .
even with all the different stellar identifications for the optical source , it is clear that must be a high mass x - ray binary ( hmxb ) .
s x - ray behavior is unique among x - ray binaries .
for instance , unlike other hmxbs , does not have persistent highly variable x - ray emission , and has a soft x - ray excess in quiescence @xcite .
the persistent soft x - ray emission can be attributed to the sgb[e ] star and the heavily absorbed harder emission may be an obscured compact object , but this is not certain , since this type of emission is extremely unusual for x - ray transients @xcite .
the flare itself peaked first in the x - ray , then in the optical , and finally at radio wavelengths @xcite , with several short , soft x - ray flares within ten days of the outburst @xcite , and rapid changes in the x - ray absorption column over the first few days @xcite , as well as years afterwards @xcite .
the radio behavior is fairly typical of x - ray transients , with an optically thick rise followed by an optically thin decay @xcite , but with no sign of further flaring activity .
an initial report of a relativistic radio jet @xcite has since been shown to be an artifact due to a bad calibrator @xcite . in this paper
we present a radio imaging study of from a few days to roughly a year after the x - ray outburst , using the national radio astronomy observatory s ( nrao ) very long baseline array ( vlba ) and the united kingdom national facility multi - element radio linked interferometer network ( merlin ) .
cccccccc 50904.92 & 1 apr .
98 & 1.28 & vlba & 1.667 & 2.8 & 3.6 & 0.23 + 50908.04 & 5 apr .
98 & 4.4 & br fd hn la mk nl sc y1 & 1.667 & 2.1 & 221.7 & 0.28 + 50908.04 & 5 apr .
98 & 4.4 & br fd hn la mk nl sc y1 & 4.987 & 2.0 & 524.6 & 0.59 + 50908.04 & 5 apr .
98 & 4.4 & br fd hn la mk nl sc y1 & 15.37 & 1.9 & 190.6 & 0.53 + 50978.74 & 14 june 98 & 75.1 & vlba + y1 & 1.667 & 1.7 & 41.5 & 0.085 + 50985.61 & 21 june 98 & 82.0 & vlba + y1 & 1.667 & 1.7 & 39.3 & 0.086 + & @xmath079 & vlba + y1 & 1.667 & 3.4 & 40.3 & 0.075 + 50996.65 & 2 july 98 & 93.0 & vlba + y1 & 1.667 & 2.1 & 36.1 & 0.084 + 51066.45 & 10 sep .
98 & 162.8 & vlba + y1 & 1.667 & 6.8 & 20.0 & 0.036 + 51209.44 & 31 jan .
99 & 305.8 & merlin & 4.994 & 10.0 & 4.3 & 0.15 + was observed on 1 april , 5 april , 14 june , 21 june , 2 july and 10 september 1998 with the full 10-element vlba with the addition of one antenna of the very large array ( vla ) , except for the 5 april 1998 observation , which was missing the pie town , kitt peak , and owens valley vlba antennas .
all of these observations were carried out using left and right circular feeds with a bandwidth of 32mhz in each polarization centered on 1.7ghz , except for 5 april 1998 , which used three bands centered on 1.7 , 5.0 and 15.4ghz .
merlin , a 6 element interferometer in great britain , observed on 31 january 1999 at 5.0ghz , with a bandwidth of 16mhz in each polarization , again using circular feeds . table 1 summarizes these observations .
the vlba observations were correlated at the vlba correlator in socorro , nm .
figure [ lc ] shows the epochs of the vlba observations in the context of the radio light - curves , taken from the green bank interferometer ( gbi ) ( rupen et al .
, in prep .
; see also @xcite ) .
the radio spectrum evolves smoothly from optically thick ( @xmath7 , @xmath8 ) two days after the x - ray peak , to optically thin ( @xmath9 ) synchrotron radiation , 7 days after the x - ray peak .
s radio emission is also , for x - ray novae , unusually long - lived , allowing it to be observed many months after the x - ray flare .
the lack of secondary radio flares , in contrast to most other x - ray novae , allows a direct link between the radio remnant and the original impulsive event .
is in fact still detected ( 2004 ) at the @xmath10mjy level with the vla ( rupen , in prep ) .
all the vlba observations were phase referenced to the strong vlbi sources j0419 + 5722 and j0422 + 5324 , while the merlin experiment was only referenced to j0419 + 5722 .
both of these sources are less than @xmath11 away from and are excellent calibrators . for the vlba observations we used a cycle time of 1 minute on j0419 + 5722 , 2.5 minutes on , then 1 minute on j0422 + 5324 .
the observations were reduced in the standard manner using the nrao s astronomical image processing system @xcite . the second epoch was phase and amplitude self - calibrated , but all other epochs were not . during the epoch the total flux smoothly varied by 10% in the 1.7 and 5.0ghz and 20% in the 15.4ghz data .
the variations were flattened by the amplitude self - calibration and did not affect the structures in the images .
the two june epochs were concatenated and imaged together .
evolves slowly enough that this is valid , and it improved the image considerably . on 1 april the vlba observed the estimated x - ray position , which was off by 11.4arcminutes . was thus at the @xmath12 power point of the vlba antennas at 1.7ghz , so all flux densities for that date
have been multiplied by 1.7 .
since these observations were phase referenced the relative positions between the high frequency the maps are accurate , although the relative positions of the 1.7ghz maps may differ by up to a few milliarcseconds ( mas ) due to the ionosphere .
also note that in these images 1mas@xmath13au ( @xmath14kpc , with @xmath15 the distance to the source ) .
a careful polarization calibration was not done , however we can say that is not polarized by more than a few percent in any epoch , probably less than that for the second epoch .
figures [ april1 ] and [ april5 ] show the images from the first and second epochs . on 1 april 1998 ,
during the radio rise , the source was very faint , and still optically thick at 1.7ghz ( see figure [ lc ] ) .
the map shows a peak of 2mjy / beam ; the faint extension to the southeast ( figure [ april1 ] ) is probably a noise spike .
may be slightly resolved ( @xmath16 ) along a position angle of @xmath17 , but this is very close to the position angle of the beam ( @xmath18 ) , and so may not be real . by 5
april the turnover frequency between optically thick and optically thin emission was about 5ghz , and the images ( figure [ april5 ] ) show a much more complicated structure . at 1.7ghz
, the source is resolved but very smooth , extending @xmath012mas in the north - south direction . at optically thin frequencies , 5 and 15ghz ,
breaks up into several condensations with a `` kidney - shaped '' structure to the south .
there are significant differences between the images made at different frequencies on 5 april 1998 : the 1.7ghz image extends further north than the 5 and 15ghz images would lead one to expect , while the western arm of the structure seen at 5ghz is not present in the 15ghz image . some of these differences could easily be instrumental .
interferometers act as spatial filters , effectively measuring different components in the fourier transform of the image .
the fourier component seen by a single baseline of length @xmath19 corresponds to a spatial frequency of @xmath20 .
thus long baselines measure rapid fluctuations of the sky brightness , while short baselines respond to more gradual changes .
the same physical interferometer , used at two different wavelengths , may respond quite differently to the identical source structure .
high - resolution interferometric images for instance may not recover large - scale structures which contain most of the source flux density : one sees the crests of the waves , but not the ocean on which they rest . for
more information see @xcite or @xcite . for on 5 april 1998 ,
the 5ghz vlba image recovers @xmath21 of the integrated flux density as measured by the vla ; while at 15ghz the vlba recovers only @xmath22 of that integrated flux density .
this strongly suggests that at least part of the 5ghz structure which is missing from the 15ghz image has been removed by the more extensive spatial filtering operating at 15ghz . to test this idea
we put the clean component model of the 5ghz data through the @xmath23 ( fourier ) coverage at 15ghz , to simulate what the 5ghz source would look like if observed at 15ghz .
this produces an image with a morphology almost identical to that actually observed at 15ghz ( see figure [ spec62 ] ) , and with a total flux which is 50% less than the same area in the 5ghz image ( i.e. , not including the western arm ) . in other words , even if its intrinsic structure were identical to that seen in the 5ghz image , the 15ghz image would not show the western arm because of the spatial filtering of the 15ghz observation .
in addition , the observed 15ghz flux density would be low by about 50% , in agreement with the actual observations .
the only notable distinction which remains is a @xmath24mjy spur to the northeast of the main emission in the 5ghz image , which is not seen at 15ghz .
the spectral index of this spur must therefore be somewhat steeper than the @xmath25 ( @xmath8 ) which characterizes the rest of the kidney - shaped structure . for comparison , the overall 5/15ghz spectral index on this date , as measured with the vla ( rupen et al .
, in prep . )
, is @xmath26 .
is significantly more extended at 1.7ghz than at the higher , optically - thin frequencies on 5 april 1998 .
this is evident in figure [ april5 ] , but is shown more clearly in the super - resolved 1.7ghz image in figure [ specind]a , where the clean component model is restored with a @xmath27mas beam .
following the same procedure as above , figure [ specind]b shows the 1.7ghz model put though the 5ghz @xmath23-coverage .
while the southern , kidney - shaped emission is fairly similar , the northern component is seen only at 1.7ghz , implying a very different spectral index .
quantitatively , the spectral index of the southern , kidney shaped component between 1.7 and 5ghz is @xmath28 , while that of the northern component must be steeper than @xmath29 .
this is most readily interpreted as an opacity gradient across the source , with the northern component being much more optically thin .
images from further vlba observations on 14 - 21 june , 2 july , and 10 september 1998 are displayed in figure [ later ] .
since does not evolve rapidly the 14 and 21 june data sets were concatenated , significantly improving the resulting image .
figure [ merlin ] shows a merlin image from 31 january 1999 .
the crosses in these images indicate the position of the peak of the 1.7ghz image from 5 april 1998 .
several results are obvious from these figures .
first , the source does not bear any resemblance to a relativistic jet .
second , has significantly expanded , from @xmath012mas on 5 april 1998 , to at least 350mas in the north - south direction on 31 january 1999 , 306 days after outburst .
third , the expansion is roughly symmetric about the center .
fourth , there is significant emission within the source s outer rim ( shell ) . finally , the shell s appearance changes significantly with time , although the position angle of the major axis is constant . in early april ,
the southern part is the brightest ; by june - july the northern portion is brighter , and the structure looks almost like a double ring . in september 1998
the source seems more nearly elliptical , and an area in the south has brightened from 1.9mjy in june to 2.7mjy in september , measured at the same position angle on the source and approximately the same area .
unfortunately , the merlin resolution is not sufficient to see any morphological details , but the west side is brighter than the east , showing that the source remains asymmetric through at least 31 jan .
it is also interesting to note that the position angle of the slightly resolved component in the first epoch ( @xmath17 ) is comparable to that of the major axis in these later images , suggesting that this position angle was established within a day of the outburst .
the radio emission in must be primarily synchrotron , from relativistic particles spiraling around strong magnetic fields .
this is based on the high brightness temperatures ( up to a few billion kelvin ) seen in the vlba images , and on the radio spectrum at high frequencies , which follows a steeper power law than allowed for thermal emission .
further , any thermal plasma hot and dense enough to produce the observed radio emission at late times would necessarily produce strong x - ray emission as well , at levels well above what is actually observed @xcite .
interestingly , the peak brightness temperature in these resolved , optically - thick images is well below the canonical @xmath30 brightness temperature limit thought to be typical of other synchrotron sources , such as extragalactic jets ( e.g. , @xcite ) and radio supernovae ( e.g. , @xcite ) .
taking the peak of the 5 april 1998 5ghz image as an example , assuming energy equipartition between relativistic electrons and the magnetic fields , and taking the ratio of proton to electron energies as @xmath31 ( as in local cosmic rays ) , the magnetic field is @xmath32 , the total energy in relativistic particles and fields is @xmath33 , and the synchrotron lifetime at 5ghz is @xmath34 .
the energy of the electrons producing synchrotron radiation at a frequency @xmath35 is then @xmath36 .
these values should be taken as indicative rather than definitive .
nevertheless it is clear that , as usual , the energy in relativistic particles and fields is not an important part of the overall energy budget the total x - ray emission in the flare for instance was @xmath37 , while the total integrated luminosity was a factor @xmath38 higher @xcite .
the rapid radio decay further shows that synchrotron losses do not dominate the observed radio light curve , another not - unexpected result . at early times
the radio remnant is optically thick , indicating substantial absorption .
the rapid changes in optical depth show that this absorption is associated with the source itself , rather than some unrelated medium .
there are two main processes which are relevant here : synchrotron self - absorption ( ssa ) by the same relativistic electrons which produce the emission , and free - free absorption by dense ionized gas along the line - of - sight .
the equipartition magnetic field derived above would produce an ssa cutoff frequency of @xmath39 ; below this frequency the emission should be optically thick , with a characteristic @xmath40 spectrum .
this is consistent with the observed 1.7 and 5ghz flux densities on 5 april 1998 , which imply a turnover frequency around 2.4ghz .
the northern component , seen only at 1.7ghz , could easily be optically thin on the same date , since its much lower flux density yields a predicted ssa turnover below 1ghz . at the same time
the emission from the main , southern region must truly be as smooth as it appears in these images : any significant clumping would lead to brightness temperatures above @xmath41 , corresponding to ssa turnover frequencies above 5ghz .
is also known to have a dense stellar wind , which could lead to substantial free - free absorption . to produce unit opacity at 3ghz
requires an emission measure ( @xmath42 ) of @xmath43 , where @xmath44 is the electron temperature , @xmath45 is the electron number density , and @xmath46 is the line - of - sight path length . taking an @xmath47 wind , assuming cosmic abundances , and taking the inner radius of the absorbing material to be 10au ( corresponding roughly to the size of the 5 april 1998 emission region ) ,
this gives a density at 10au of @xmath48 .
for comparison the modeled density for the wind in the sgb[e ] star hd87643 at 10 stellar radii is @xmath49g@xmath50 and @xmath51g@xmath50 for the hot diffuse and cold dense winds , respectively @xcite .
free - free absorption could thus be quite significant , depending on the relative orientation of the winds and the radio - emitting region , and on the temperature and ionization of the winds . in sum
, synchrotron self - absorption must be important in the early radio remnant , but free - free absorption may also play a role . a detailed analysis of the multi - frequency radio light curve may break this degeneracy . even without any absorption
, the radio remnant shows significant spectral variations : the northern component seen at 1.7ghz on 5 april 1998 must for instance have an intrinsic emission spectrum ( @xmath52 or steeper ) which is much steeper than the rest of the source ( @xmath53 ) .
such extreme variations have not been seen in other shell - like sources , such as radio supernovae ( bietenholz et al .
, priv.comm . ) and supernova remnants ( e.g. , @xcite ) .
the explanation may lie in the very complex environment around , which could plausibly lead to very different densities and magnetic fields in different parts of the circumstellar medium .
the basic appearance of s radio remnant ( edge - brightened , clumpy , shell - like ) strongly suggests a shock plowing into a dense circumstellar medium .
the radio emission from all other x - ray transients has been either unresolved , or consistent with a highly collimated one- or two - sided jet ; but there is certainly no sign of a jet in these images , and the continued brightness of the remnant despite substantial expansion requires on - going particle acceleration .
the relative brightness of various features changes substantially from epoch to epoch , with some parts of the remnant even brightening despite the overall decay ; but the overall shape remains fairly constant ( see below ) .
the basic implication is that we are seeing the evolving results of an explosion inside a dense , clumpy environment , much like a radio supernova .
the changing appearance then corresponds to changes in the local shock strength , the particle acceleration efficiency , or the synchrotron emissivity ; these in turn are presumably functions of the local particle density and magnetic field strength and orientation . even in our last image ,
the remnant shock is only @xmath54mas ( @xmath55au ) across , and is thus still running into material dominated by s stellar wind .
the highly asymmetric and clumpy radio morphology probably implies a similarly anisotropic and inhomogeneous circumstellar medium ( csm ) , which seems more likely than a severely asymmetric explosion or magnetic field , and is consistent with other indications of a highly anisotropic wind @xcite .
the idea of a dense surrounding medium is also supported by x - ray observations of in quiescence , which show strong and highly variable absorption @xcite , and by s strong infrared excess @xcite . in this picture
the changing appearance of the remnant reveals the variations in the density of the csm .
the initial bright kidney shaped structure to the south ( figure [ april5]b and c ) was presumably due to a particularly dense clump on that side , which has been over - run by the time of the june - july images ( figure [ later]a and b ) ; while the initially fainter emission to the north indicates a lack of dense material on that side . @xcite and @xcite suggest that the outflow from consists of a slow ( @xmath56 ) , dense equatorial wind , and a fast ( @xmath57 ) , less dense polar wind .
the shock seen in the radio emission between 2july1998 and 10september1998 then traces out 250/8200 years of stellar mass loss , during which time the density of the stellar wind changed from most dense in the north - west to most dense in the south - west ; while the similarity between the june and july maps shows that the wind did not vary dramatically on shorter timescales ( 75/2500 years ) .
significant stellar wind variations are not unprecedented
. optical images of icr+10216 , for instance , reveal discrete incomplete rings of dust , which imply that the mass loss rate of the star changes on a hundred year timescale , and is not spherically symmetric @xcite . despite these changes in brightness across the source ,
the overall structure is quite similar from epoch to epoch : that is , the brightest regions move outward at the same rate as the dim ones .
this suggests that the radio synchrotron emission simply serves to ` light up ' the edges of the shock , while remaining energetically unimportant . by itself
this is not so surprising , as the luminosities are so low , but this further implies that the particle acceleration , and whatever causes it to vary around the source , is also not dynamically important .
this is a common result for shocks .
the radio remnant of sn1993j , for instance , exhibits changing 3:1 brightness asymmetries around a shell which decelerates but remains highly circular @xcite and the supernova remnant sn1006 has huge brightness asymmetries around a similarly circular rim @xcite . as with most astronomical imaging ,
these data show a two - dimensional projection of a three - dimensional object .
the radio remnant is clearly elongated , so the intrinsic structure must also be asymmetric , either due to an off - center explosion , or to an asymmetry in the surrounding medium .
one puzzle is that the source retains a similar appearance while slowing down considerably ( see below ) .
several possible geometries , which could explain the overall appearance , are shown in figure [ scheme ] .
an ellipsoid could explain the outer shape of the remnant , with the brighter regions in the middle caused by some minor asymmetry in the particle acceleration / emission process along the line - of - sight to the center .
however , it is difficult to retain the axis ratio of such an ellipsoid over time .
any three - dimensional structure confined by denser material around the ` waist ' would be expected to become more and more elongated over time , as the material in the waist should decelerate more rapidly than that at the poles .
edge - brightened cone rims could result from a shock lighting the densest regions of a dense wind , as in some planetary nebulae , while a pair of spherical bubbles might arise in an explosion in a highly asymmetric environment .
figure [ fits ] shows a tilted circular ring and a two - sided tilted cone that were fit to the data ; note that the rims of the cone both move outward and expand with time . in both cases the expansion rate is @xmath58 , as discussed in the next section .
the angle between the normal to the titled ring and the line - of - sight is @xmath59 , while the angle between the cone axis and the line - of - sight is @xmath60 .
these fits are illustrative rather than unique .
the two - sided cone seems to fit marginally better as it takes into account the waist of the nebula .
the results for a pair of spherical bubbles would be similar to those for the two - sided cone .
the expansion rate of can be fit regardless of the model used to explain the morphology , assuming that the source remains self - similar throughout .
figure [ exp ] shows the results of fitting an ellipse to each image ; the axis ratio is constant at 1.9:1 , and this plot shows the expansion of the major axis as a function of time .
the shock is clearly decelerating , presumably because of material swept up from the wind as the shock expands .
the average radial expansion rate of the major axis between an assumed explosion date of 31.64 march 1998 ( corresponding to the first x - ray detection ) and 5 april 1998 is @xmath61 @xmath62 .
this means during the 5 april 1998 observations the source expanded by @xmath63 . since the beam is bigger than this for the 1.7 and 5ghz images this would have no effect .
however @xmath63 is comparable to the 15ghz beam and therefore the motion may have smeared the image slightly in the direction of motion .
this is probably insignificant though , since the structures in this image are @xmath64 .
between the first x - ray detection and the merlin image ( january 1999 ) the _ average _ expansion is @xmath65 @xmath66 .
a power - law gives a reasonable fit to the expansion data : @xmath67 , where @xmath2 is the major axis in mas , @xmath3 is the modified julian date ( mjd ) , and the error bars are @xmath68 assuming normally - distributed errors and forcing @xmath69 .
the implied radial velocity is @xmath70 . while only poorly determined , the explosion date is consistent with the beginning of the x - ray flare ( on mjd 50903.64 ) , and the fit is consistent with being marginally resolved in the first epoch ( @xmath71 ) .
this expansion rate ( @xmath72 ) is much faster in than the classic sedov phase seen in supernovae remnants ( @xmath73 ) , but is consistent with young supernovae ( e.g. , for sn1993j @xmath74 [ @xcite ] ) . given the density of s circumstellar medium , the initial impulsive energy injection must have been quite strong to produce such a fast shock .
one quantitative estimate comes from the changing x - ray absorption column seen at early times @xcite .
this implies column densities @xmath75 of a few @xmath76 which decline by more than a factor of ten over the first two days .
it seems reasonable to associate this decline with the forcible removal of material by the shock seen in the radio images . by the time the x - ray absorption has decayed away on 2 april @xcite the shock was @xmath77 @xmath78 across , according to the above fit .
the total number of atoms swept up was @xmath79 @xmath80 , giving an average density of @xmath81 , or @xmath82 , assuming an average mass per atom of 1.4 times the mass of a proton .
these values are well within the range implied by spectroscopy during the outburst ( @xmath83 to @xmath84 : @xcite ) , and not inconsistent with models of the winds in sgb[e ] stars ( e.g. , @xcite ) .
the total mass swept up by the shock was then of order @xmath85 , giving a corresponding kinetic energy ( at @xmath86 ) of @xmath87 .
this number is very uncertain , given all the underlying assumptions . even so it is clear that the kinetic energy in the shock could easily match or exceed the integrated luminosity of the outburst across all observed wavelengths ( @xmath88 , @xcite ) .
the continued slow deceleration of the shock can be used to place constraints on the distribution of the circumstellar material via the shock jump conditions , in a manner analogous to @xcite s treatment of the expansion of radio supernovae .
actually what is derived is a relationship between the density structure of the outer regions of the ejecta , and that of the csm . here
the ` ejecta ' should be thought of as the material swept up by the time of the first observation ; while this is the stellar envelope in the case of a supernova , for it is probably more appropriate to think of dense material surrounding ( accreting onto ? ) the compact object .
the shock between ejecta with outer density profile @xmath89 and a csm with density profile @xmath90 will expand at a rate @xmath91 , with @xmath92 .
the observed @xmath93 is well within the range observed for radio supernovae ( e.g. , @xcite ) , and suggests @xmath94 for an @xmath95 wind profile . more generally one can argue that the observed deceleration implies that the shock has swept up a significant amount of mass , compared to that of the original ejecta .
the factor two drop in velocity at late times implies a factor two to four increase in the mass over the same interval , assuming momentum or energy conservation , respectively .
data taken before the april 1998 outburst showed a stable spectrum dominated by symmetric , single - peaked hydrogen balmer lines , with wings extending to @xmath96 @xcite . during the outburst both and lines became much broader and asymmetric , with line wings extending up to @xmath97 in h@xmath98 and on 3 april 1998 , and perhaps out to @xmath99 in the blue wing of h@xmath98 @xcite .
@xcite also see @xmath100 line wings in h and he on 9 april 1998 . while these broad wings faded fairly quickly @xcite , an ultraviolet spectrum taken two years after the outburst showed p cygni profiles with absorption wings extending to @xmath101 @xcite .
it seems natural to associate these broad lines with the outflow imaged at radio wavelengths .
the proper motions from our images correspond to a radial expansion of the major axis at @xmath102 .
the shock velocity was thus @xmath103 on 3 april , and @xmath104 on 9 april 1998 ; extrapolating to two years after the outburst gives an expected velocity of @xmath105 .
the corresponding expansion velocities of the minor axis of the fitted ellipse are 60% of these values .
there are many uncertainties in comparing the radio proper motions with the optical doppler shifts .
first , the source is clearly asymmetric , both from the radio images and from the optical line shapes at early times , which makes the comparison of radial with transverse velocities somewhat suspect .
second , while the radio emission traces the motions of the outer ejecta , the optical lines could be sampling very different regions , with different kinematics .
this is often seen in supernovae ( e.g. , @xcite ) , and is clearly the case for at later times , when no substantial line wings are seen .
given the difficulties the rough agreement between optical and radio velocities is remarkable , and suggests that the distance to the source is within a factor two of the nominal 5kpc we have adopted .
there are some similarities between the radio morphology of and the radio outflows / nebulae observed in other stars ( e.g. , ag pegasi [ @xcite ] and hm sagittiae [ @xcite ] ) , as well as optical outflows around planetary nebulae ( pne ) @xcite and the high mass star eta carinae @xcite .
the hourglass and double ring structures seen in many pne is thought to be caused by the density structure of the circumstellar material into which the ejected material expands @xcite ; this is very similar to our picture of s remnant .
however , the optical emission from pne has a very different physical origin from the radio synchrotron seen in , as does the radio emission from most stellar outflows ( thermal free - free radiation ) .
the expansion rate in is also much larger than the 10km / s to 1000km / s seen in other stars ( e.g. , @xcite ) . the overall structure of the radio remnant ( a clumpy shell ) ,
its initial expansion speed ( @xmath106 ) , and its gradual deceleration are all very similar to what is seen in radio supernovae ( e.g. , sn1993j : @xcite ) .
there are however two basic differences .
first , the radio shell in is decidedly asymmetric , while the few resolved radio supernovae are basically circular .
supernova remnants on the other hand are often asymmetric , which ( as in ) is most readily interpreted as the result of a shock running into a severely anisotropic medium .
second , while radio supernovae and indeed , virtually all resolved , optically - thin radio sources have rather steep radio spectra ( @xmath107 to @xmath29 ) , retains a much flatter @xmath108 for many months .
since the optically - thin spectral index is a direct consequence of the energy distribution of the relativistic electrons , this requires a very unusual ( but also quite stable ) particle acceleration process . while radio emission is often associated with x - ray transients and binaries ( e.g. , @xcite ) ,
is unique in showing a slowly - expanding , two - dimensional remnant . in all other cases ,
any resolved radio emission is dominated either by separable ejecta , or by a collimated jet , often moving at relativistic speeds ( e.g. , @xcite ) .
this has led to the wide - spread assumption that radio emission from x - ray binaries always indicates the presence of a relativistic jet , and many models now take radio emission as the best unambiguous tracer of jet emission ( e.g. , @xcite )
. throws this entire paradigm into doubt .
does this have wider implications for x - ray binaries in general ? or is truly unique ?
one possibility is that did indeed produce a fast jet , but this jet was quickly smothered by the dense surrounding material . in this model
the energy in the jet was converted into a less directed outflow , which we see as a shock moving out into the csm .
this is similar to the failed - jet model for supernova explosions in collapsars ( e.g. , @xcite ) .
quantitatively , one expects a relativistic jet to slow catastrophically when it has swept up an equivalent rest mass energy ( e.g. , @xcite ) : @xmath109 where @xmath110 is the energy in the jet , and @xmath111 is the average density within the radius @xmath112 at which the jet slows catastrophically .
this equation is appropriate to a jet which is stopped quickly enough that lateral expansion can rapidly isotropize the energy . for , we know that the remnant is far from jet - like by a radius of 6mas ( @xmath113 ) , and according to the arguments in [ sec : exp ] the energy in the shock ( in this model derived entirely from the original jets )
is of order @xmath87 .
this gives a lower limit to the mean density required to stop the jets of @xmath114 , well within the plausible range for the wind of a sgb[e ] star .
it is interesting to note that simulations of `` stalled '' jets in the context of @xmath6-ray bursts predict morphologies very similar to what we see in ( see particularly the @xmath115-inclination images in figures 8 and 9 of @xcite ) . the question arises as to why other jets are not similarly smothered .
most other x - ray transients live in relatively pristine environments , where the jet might more easily escape . in binaries with substantial circumstellar material
the accretion tends to be more continuous ( e.g. , cyg x@xmath1163 and ss433 ) , and the resulting jets are also ` on ' more often , and are thus perhaps capable of sweeping away the csm .
this fits in with @xcite s picture of the outburst , in which a compact object in a highly eccentric long - period ( @xmath117 ) orbit moves through the dense equatorial wind of its companion star ; the brief period of super - critical accretion , when the compact object is close to the companion , leads to the x - ray flare and the expanding radio shock front .
a long - period orbit is required to explain the lack of previous known flares , which would have been obvious in the optical to historical astronomers . given that s radio remnant is clearly dominated by a shock , it is natural to ask whether the emission at other wavelengths could also be associated with that shock , rather than with the accretion disk .
extrapolation of the radio synchrotron spectrum to other wavelengths shows that the expected synchrotron emission is far below that observed during the outburst @xcite .
similarly the thermal emission produced by the shock is likely to be quite modest , for three reasons .
first , the shock has not had enough time to become radiative , for any reasonable densities .
second , the characteristic temperature of the forward shock ( @xmath118 for a 10,000km / s shock ) ( a ) lies outside the passband of most available instruments , and ( b ) puts the shocked gas in a regime where cooling is quite slow .
third , while the reverse shock should be slower and hence lead to lower and more easily observable temperatures , the resulting fluxes are likely to be very low ; scaling from the radio / x - ray ratios in observed supernovae for instance ( e.g. , @xcite ) the expected emission is orders of magnitude below the observed flux during the outburst .
the observed soft x - rays , in this as in other x - ray transients , probably arise in and around the accretion disk rather than in the shock .
thus the only obvious signatures of the shock outside the radio remnant are the optical and uv spectral lines .
some of these are doppler broadened by a thousand or more km / s ; others evolve very rapidly , requiring very high densities , as might result from a shock @xcite .
recent spectra ( 2001 - 2002 ) still show many spectral lines at very different strengths than their very stable pre - outburst levels @xcite , which suggests some continued disturbance around the binary .
none of these is really a ` smoking gun , ' indicating that , for an unresolved x - ray transient , a shock and a freely expanding jet may not be distinguishable
. it does not however seem likely that the radio emission from many other unresolved x - ray binaries can be fully explained by shocks similar to that seen here .
s radio light curve is unusual in showing a single outburst followed by a smooth , steep - spectrum ( @xmath119 ) decay .
outbursts from other sources often lead to multiple radio events , suggesting strong on - going activity more readily associated with the central source than with a single expanding shock wave . there could be a series of shocks of course , but that requires re - furbishing the surrounding medium on rather fast timescales ( days to weeks ) .
on the other hand , this single shock evacuating the surrounding medium may explain why there was only one explosive event , i.e. , the shock got rid of any additional material that could have been accreted by the compact object , and the compact object moved away from the high density region before the stellar wind could replenish the material .
the quasi - steady radio emission seen in the low / hard state is also very different from s optically - thin remnant , and requires a very small and continuously replenished radio source . given these indications of very different power sources , it is quite surprising that lies nicely on the loose x - ray / radio correlation found for x - ray binaries @xcite , with a scatter typical of other x - ray transients .
it is far from obvious why direct synchrotron emission from jets should have the same efficiency as particle acceleration in an expanding shock .
we have presented evidence that the radio remnant of s 1998 outburst takes the form of a clumpy , decelerating shell , strongly suggesting an expanding shock front , rather than the relativistic jets seen in many other x - ray transients .
we suggest that the reason for this distinction lies in s very dense stellar wind , which could easily convert even quite energetic jets to a quasi - spherical outflow .
the very broad spectral lines seen in optical emission early on , and in the ultraviolet several months later , seem to be the only other unambiguous signatures of this outflow .
associating the radio outflow with those doppler shifts suggests that the adopted distance of 5kpc is probably good to a factor of two . assuming the changing x - ray absorption at early times is due to circumstellar material removed from the line - of - sight by the radio shock
, we estimate the shock energy as @xmath87 , comparable to the total integrated luminosity of the outburst .
if the smothered jet model is correct , this represents one of the very few measurements of a jet s total kinetic energy .
the relativistic jets seen in other x - ray binaries and transients must either be much more powerful , or be expanding through much less dense material .
the radio light curves of most unresolved sources suggest continued emission from a small region , so it also seems unlikely that extended shocks dominate the radio emission for these sources
. seems to be related to other x - ray transients , as collapsar supernovae are related to @xmath6-ray bursts .
this paper would not have existed without robert hjellming , who provided the drive to observe this source with the vlba even before we knew it was a strong radio source .
he also edited the first 40 days of the gbi data .
we are further indebted to vivek dhawan , kristy dyer , dale frail , and rob hynes for useful discussions .
the vla and the vlba are facilities of nrao which is operated by associated universities , inc . under cooperative agreement with the national science foundation .
the gbi is a facility of the national science foundation operated by the national radio astronomy observatory ( nrao ) in support of nasa high energy astrophysics programs .
a.j.m . acknowledges support from the european commission s tmr / lsf programme ( contract no .
erb - fmge - ct95 - 0012 ) .
basic research in radio astronomy at the naval research laboratory is funded by the office of naval research .
this research has made use of the simbad database , operated at cds , strasbourg , france , and nasa s astrophysics data system . | we present radio imaging observations of the 1998 outburst of the peculiar emission line star , taken @xmath01 , 4 , 75 , 82 , 93 , 163 , and 306 days after the beginning of the 31.64 march 1998 x - ray flare .
the first two epochs show a resolved but compact ( no larger than 12 milliarcseconds ) radio source which becomes optically thin at frequencies higher than 5ghz .
the spectrum and brightness temperatures are consistent with synchrotron self - absorption , although free - free absorption may also play a role .
the later images show a large ( 120 - 350milliarcseconds ) oval - shaped or double ring remnant .
the radio spectrum combined with the high brightness temperature indicates that the emission is synchrotron , while the morphology suggests that this is powered by a decelerating shock moving through dense circumstellar material produced by a strong stellar wind .
the radio images of are equally well fit by an expanding ellipsoid or two expanding rings ; the former gives @xmath1 , with @xmath2 the major axis in mas and @xmath3 the modified julian date ( mjd ) .
the corresponding expansion speed in the plane of the sky was @xmath4 over the first few days ( for an assumed distance of 5kpc ) , slowing by a factor @xmath5 by the time of the last observation almost a year later .
the radio emission from all other x - ray binary transients has either been unresolved , or taken the form of highly collimated relativistic jets .
we suggest that represents a rare case where these jets were smothered early on by the unusually dense circumstellar medium . in this model
is the analogue to extragalactic supernovae formed by the collapsar mechanism , while the more usual x - ray binaries with relativistic jets are analogous to the jets which escape those supernovae to form a subset of @xmath6-ray bursts . |
the multifractal detrended fluctuation analysis ( mfdfa ) @xcite became in last ten years the major technique used for studying the multifractal properties in complex systems and in time series .
more then 500 papers discusses mfdfa issues in variety of problems related with complexity ( see , e.g. , @xcite@xcite ) . the central role in mfdfa
is played by the @xmath1-th moment of fluctuation function ( @xmath3 ) defined as @xcite @xmath4^{q/2}\right\}^{1/q}\ ] ] for @xmath5 , and @xmath6\right\}\ ] ] for @xmath7 , where @xmath8 and @xmath9 ( @xmath10 ) are data in series , @xmath11 is the size of window box in which detrending is performed , while @xmath12 is the polynomial trend subtracted for @xmath13-th data in @xmath14-th window box ( @xmath15 ) .
the observed scaling law @xmath16 is crucial within mfdfa to estimate the multifractal properties of a signal given by the profile of generalized hurst exponents @xmath0 .
the strength of multifractality present in data is usually studied as a spread @xmath17 @xmath18 of some edge values of generalized hurst exponent @xmath19 calculated at the fixed negative ( @xmath20 ) and positive ( @xmath21 ) moments respectively ( @xmath22 ) . in the case of stationary series
, @xmath0 is proven to be a monotonically decreasing function and therefore @xmath23 @xcite . however , many examples shown recently in literature for synthetic data @xcite as well as for realistic signals @xcite reveal that multifractal properties may lay far outside this simplest scenario and the spread @xmath24 , defined as the difference of generalized hurst exponents calculated for small negative and large positive fluctuation moments , is not indicative in these cases . a non - monotonic behavior of @xmath0 profile is observed for instance in nonstationary data what leads to twisted @xmath25 multifractal spectrum @xcite if the hlder description of multifractality is adopted with a help of legendre transform @xcite . in the case of nonstationary data with periodicity ,
white or color noise added ( see , e.g. , @xcite ) one may see domains where @xmath0 is either increasing or decreasing with @xmath1 , local maxima in @xmath0 are formed or even @xmath26 suggesting that big fluctuations may appear more often than small ones .
it is contrary to observations in stationary data @xcite where it should be the other way round .
there are examples of realistic time series where the multifractal profile , despite small and insignificant @xmath17 spread , significantly expands in the central part for moment orders @xmath27 .
one finds such behavior for instance for nonlinearly transformed financial data containing extreme events ( see , e.g. , @xcite ) .
the similar spread @xmath17 may correspond also to significantly different characteristics of the @xmath0 profile in the center for @xmath27 .
the above results convince that a traditional measurement of multifractality within mfdfa , based entirely on the @xmath17 spread may be misleading . therefore we propose in this article an alternative measure of multifractal properties which takes into account the behavior of the multifractal profile @xmath0 as a whole for all @xmath1-th order moments , not limited only to the edge values @xmath19 found at @xmath28 .
as pointed above , the new measure should be sensitive to behavior of the whole @xmath0 profile in the wide range of @xmath1 moments ( @xmath29 ) and should not neglect the multiscaling properties within this range of @xmath1 s , contrary to the standard measure based on the spread @xmath17 alone which takes into account just the edge values of the generalized hurst exponents @xmath30 .
in the simplest case of multifractal data with the abandoned effects of so called multifractal bias , generally discussed in @xcite , a natural extension of @xmath17 can be defined as a cumulated distance between the multifractal profile @xmath0 and the main hurst exponent value @xmath31 for considered series .
one may write then a new measure @xmath32 in the integral form @xmath33 it turns out that @xmath34 has an interesting interpretation of the scaling exponent in the new power law linking the geometric mean of detrended @xmath1-th order fluctuations @xmath35 of the signal for all @xmath1 values with the size @xmath11 of the time window .
indeed , consider the normalized detrended @xmath1-th moment of fluctuation @xmath36 defined as follows @xmath37 where @xmath38 is the standard deviation of detrended fluctuation ( @xmath39 ) in time window of length @xmath11 .
using the power law from eq.(4 ) we may write the relationship @xmath40 where @xmath41 is some function generally dependent on @xmath1 but independent on @xmath11 . hence , using definition from eq.(6 ) @xmath42 replacing @xmath43 by @xmath44 and @xmath1 by @xmath45 , where @xmath46 , the first integral in rhs of eq.(9 ) can be approximated by a discrete sum @xmath47^{\frac{1}{n}}\ ] ] comparing eqs(9 ) and ( 10 ) , one gets in the continuous limit @xmath48 @xmath49^{\frac{1}{n } } + \int_{-q}^{q}c(q)\,\mathrm{d}q\ ] ] the latter relation builds after simple rearrangement the postulated power law @xmath50^{\frac{1}{n } } \sim \tau^{\delta^{(2)}_h}\ ] ] note that the standard spread of generalized hurst exponents values used in mfdfa to determine the multifractal content can also be interpreted in the above context as the scaling exponent in the power law @xmath51 which links together @xmath11 with the ratio of the smallest ( @xmath52 ) and the biggest ( @xmath53 ) detrended @xmath1 deformed fluctuations .
hence , the scaling in eq.(12 ) related with geometric mean of all detrended fluctuations @xmath54 , can be regarded as the modification of mfdfa method we call gmfdfa ( geometric mean mfdfa ) .
we will test the behavior of the new measure analyzing first the synthetic multifractal signals with apriori well known multiscaling properties .
they were generated from the binomial random cascade algorithm @xcite for the almost complete range of @xmath55 parameter describing the richness of multifractal content in synthetic signal ( @xmath56 ) with a step @xmath57 .
[ delta_vs_a ] shows the plot of analytical expectation @xmath58 in this model obtained from the relation @xcite @xmath59 and compared with two numerical findings , respectively for the standard @xmath17 and `` new '' @xmath60 measure , calculated for @xmath61 from numerical simulation of @xmath0 profile .
these plots prove that in a case of synthetic data new measure of multifractal content is qualitatively and quantitatively indistinguishable from the standard measure based on the spread @xmath17 of the edges @xmath62 .
the simple recipe for the new measure will have to be modified if so called multifractal bias effects are to be included in analysis @xcite .
let us recall here main issues connected with such multifractal bias .
it is defined as the continuous set of profiles @xmath0 calculated for a given system or signal , which do not correspond to multifractal properties of this system at some confidence level , despite @xmath63 .
such bias can be seen particularly in short signals where accidental fluctuations , not related with multifractal properties , contribute to fluctuation function and dominate over fluctuations having their origin in multifractal properties of the signal due to small statistics of short data .
it raises the @xmath64 edge of the multifractal spread . on the other hand ,
the influence of large fluctuations in short data set can be suppressed since mfdfa algorithm may treat such fluctuations as longer trends and eliminate them .
in both cases the observed @xmath17 spread is increased comparing with the corresponding value obtained for much longer data and grows with persistency in signal @xcite .
another source of multifractal bias are nonlinear transformations performed on data which introduce always additional registered multifractal effect .
the level of multifractal bias in series can be calculated or simulated with a number of semi - analytic formulas provided in @xcite .
note that deliberately we do not mix this bias effect here with so called spurious multifractality @xcite caused by various external contaminations in signal or nonstationarities coming , e.g. , from the influence of noise present in data .
when taking multifractal bias into account , one has to generalize the formula in eq.(6 ) , replacing @xmath65 in there by the corresponding multifractal bias thresholds @xmath66 at the assumed confidence level ( @xmath67 throughout this article ) . here
@xmath68 are respectively the upper ( @xmath69 ) and the lower ( @xmath70 ) edge of the bias area ( ribbon ) explained in fig .
[ artistic ] ( and shown also in figs .
[ cascades_withb],[sp500][nasdaq ] ) .
simultaneously , @xmath0 profile in eq .
( 6 ) has to be replaced by the observed ( registered ) values of the multifractal profile @xmath71 coming from direct application of eq.(4 ) in mfdfa .
let us define a distance function @xmath72 as @xmath73 the modified measure @xmath74 in the case of multifractal bias present in signal can be taken as the average distance @xmath72 @xmath75 and is presented as a green area in fig .
[ artistic ] . a new measure of multifractality @xmath74 generalizes the multifractal spread @xmath17 in such a way , that it can be used for particular profiles @xmath71 even if the standard @xmath17 measure could not be applied ( e.g. , when @xmath76 or @xmath77 ) .
one always gets @xmath78 describing multiscaling properties in a system for the whole range of fluctuation moments .
the result @xmath79 means that the observed multifractal profile entirely lies in a biased ribbon domain for all moment orders @xmath1 and hence there are no expected real multifractal effects even if @xmath80 .
note , however , that the maximum value of @xmath21 at which the new measure is calculated can not be too high , since then the contribution to the profile @xmath0 with high volatility may be dominated by domains where @xmath81 .
in order to present the qualities of proposed measure @xmath74 let us start with analysis of synthetic signals obtained from binomial cascade model , additionally , taking into account the multifractal bias .
the calculated multifractal profiles @xmath0 , together with ribbon - like areas representing the multifractal bias are gathered in fig .
[ cascades_withb ] for two values of cascade parameter @xmath82 and @xmath83 ( @xmath84 ) .
the values obtained for three measures applied to examples from fig . [ cascades_withb ] : observed ( biased ) spread @xmath85 , unbiased spread @xmath86 and proposed measure @xmath74 are presented in fig . [ cascades - barplots ]
. the actual values of multifractal measures corresponding to this figure are collected in table .
[ cascades - table ] .
note , that in the traditional measurement of the multifractal spread ( see fig .
[ artistic ] ) , when @xmath87 and @xmath88 we have the following relations in terms of the edge @xmath89 values : @xmath90 @xmath91 where @xmath92 are the observed ( biased ) edge values of the multifractal profile and @xmath93 are increments above ( below ) the multifractal bias level .
the latter ones are assumed to be null if the edges @xmath94 satisfy @xmath95 .
therefore , one gets @xmath96 where @xmath97 is the bias level and @xmath98 is the unbiased spread of generalized hurst exponents . in case of synthetic multifractals , the new measure @xmath74 gives results similar to unbiased multifractal spread @xmath86 as seen from tabularized values in table [ cascades - table ] .
both measures ( @xmath86 and @xmath74 ) subtract the influence of finite size ( fse ) and nonlinear transformation effects . however , when a real complex system data are considered significant difference between new and standard measure can be observed .
we present an evolution of three stock exchange indices : s&p500 , nikkei225 and nasdaq in figs .
[ sp500 ] , [ nikkei225 ] and [ nasdaq ] in two adjacent time windows .
investigated examples are chosen in such a way , that each right hand side window contains an extreme event , that is : `` black monday '' of 19 .
october 1987 for s&p500 , fall crash in 2008 for nikkei225 and the crash of 30 .
august 1995 for nasdaq , while the left hand side window corresponds to proceeding `` quiet '' period on the stock market .
all data are investigated for primary price increments @xmath99 and their nonlinear transformations : absolute price increments @xmath100 , quadratic increments @xmath101 , absolute returns @xmath102 and moving volatilities @xmath103 , @xmath104^{\frac{1}{2}}$ ] , where @xmath105 and @xmath106 transaction days
. the properties of multifractal profiles @xmath0 and multifractal bias coming from fse and nonlinear transformation effects ( red ribbon - like area ) are presented in the first and last column in each figure .
the quantitative comparison of multifractal measures based on @xmath17 and @xmath74 exponents calculated from these plots are collected in tables 24 and plotted in figs .
[ sp500-barplots][nasdaq - barplots ] .
clear differences in multifractal features are visible between quiet and more dynamical periods ( differences in shapes of @xmath0 profiles ) , which makes them excellent candidates to confront the qualities of three measures @xmath85 , @xmath86 and the proposed one @xmath74 .
no qualitative difference between both periods is visible for primary data of nikkei225 index , when the standard measure @xmath86 is applied . s&p500 and nasdaq behave the other way round one observes richer multifractal structure for them in dynamical period . for nonlinear transformations
a strong non - monotonicity is mostly visible in dynamical period of all considered market indices leading in some cases ( @xmath100 and @xmath102 for nikkei225 ) to inversed behavior , where @xmath107 ( here @xmath61 ) .
thus @xmath108 and the traditional measure is not indicative in such cases .
the unbiased measure @xmath86 for @xmath100 and @xmath102 of dynamical period for s&p500 , nikkei225 and nasdaq and @xmath101 just in case of nikkei225 index causes also interpretation problems ( @xmath109 ) .
the above problematic situations are marked as cross - points in figs .
8 , 9 , 10 .
unexpectedly , moving volatile transformations @xmath110 and @xmath111 for dynamical period of nikkei225 index , show complete lack of unbiased multifractal effects contrary to other considered indices . the new measure ( see bottom panel in figs .
8 , 9 , 10 ) maintains validity of conclusions stated about multifractal features found in traditional manner for primary series @xmath99 .
but simultaneously , all prior problems with interpretation of multifractal properties of transformed data disappear .
the interesting results are seen for two leading world indices s&p500 and nasdaq , particularly in dynamical zones .
the standard measure could not provide any quantitative result to be compared with the corresponding behavior in `` quiet '' zone ( see figs .
[ sp500-barplots][nasdaq - barplots ] , top panels ) .
the new measure makes it possible ( see figs .
[ sp500-barplots][nasdaq - barplots ] , bottom panels ) .
moreover , one clearly notices that the index and its nonlinear transformations exhibit much richer multifractal content in a dynamical periods , then in the `` quiet '' zone .
nikkei index is an exception , but also here a comparison between multifractal behaviors in two domains is possible only within the newly introduced measure ( see fig . [ nikkei225-barplots ] ) .
this paper proposed the new method of measuring multifractal features in time series .
the method is based on global analysis of generalized hurst exponents @xmath0 . in the simplest case ,
when multifractal bias is not considered , the new measure is defined as cumulated distance of the generalized hurst exponents @xmath0 from the main hurst exponent @xmath112 , since @xmath112 is the only scaling exponent value one can refer to in monofractal case .
the generalisation to the case with multifractal bias from fse and nonlinear transformations was also provided .
an example of synthetically generated multifractals with binomial cascades shows , that the new measure reproduces the standard spread of multifractal profile @xmath17 , both before and after transformation .
however , when multifractal bias is taken into account , the new measure seems to be much more resistant to disturbances coming from different sources .
we have presented the examples of real signals drawn from financial markets .
these examples prove , that the new technique of measuring multi - scaling features in time series is useful , especially in very dynamical periods of complex system evolution , when simpler predecessor ( @xmath17 ) ceases to work properly ( see transformations @xmath100
, @xmath101 @xmath102 , in right panels of figs .
[ sp500-barplots ] and [ nikkei225-barplots ] and transformations @xmath100 , @xmath102 in right panel of fig .
[ nasdaq - barplots ] ) . obtained results seem to confirm , that whenever standard methods of detecting multifractality within mfdfa face problems , the newly proposed approach based on the scaling law from eq .
( 12 ) detects it successfully .
( simulated ) , @xmath113 ( theoretically estimated from eq .
( 14 ) ) and the newly proposed @xmath34 shown for synthetic series of length @xmath84 generated from random binomial cascade algorithm for different values of parameter @xmath55 ] @xmath115 ( black dots ) and the corresponding multifractal bias ( red area ) calculated for synthetic series generated from random binomial cascade algorithm for @xmath82 and @xmath83 first row shows result for primary data in series @xmath116 , while consecutive rows present the results of following nonlinear transformations : @xmath100 , @xmath101 , @xmath102 , @xmath110 , @xmath111.,title="fig : " ] profiles are placed correspondingly in left and right panels for primary and transformed data .
each profile is supplemented by related multifractal bias ( red ) .
values of series increments ale marked in magenta color . ] .
cross - point in top plots correspond to cases , when traditional multifractal profile spreads @xmath85 or @xmath86 causes problems with interpretation ( @xmath117 for @xmath85 or @xmath118 or @xmath119 for @xmath86 ) .
see also fig .
[ artistic ] .
, title="fig:",scaledwidth=49.0% ] .
cross - point in top plots correspond to cases , when traditional multifractal profile spreads @xmath85 or @xmath86 causes problems with interpretation ( @xmath117 for @xmath85 or @xmath118 or @xmath119 for @xmath86 ) .
see also fig .
[ artistic ] .
, title="fig:",scaledwidth=49.0% ] 50 j. w. kantelhardt , s. a. zschiegner , e. koscielny - bunde , s. havlin , a. bunde , h. e. stanley , physica a 316 ( 2002 ) 87 l. telesca , v. lapenna , m. macchiato , physica a 354 ( 2005 ) 629 l. telesca , v. lapenna , tectonophysics 423 ( 2006 ) 115 m. s. movahed , f. ghasemi , s. rahvar , m. r. r. tabar , phys .
e 84 ( 2011 ) 021103 p.
h. figueiredo , e. nogueira jr .
, m. a. moret , s. coutinho , physica a 389 ( 2010 ) 2090 s. dutta , j. stat .
12 ( 2010 ) p12021 i. t. pedron , journal of physics : conference series 246 ( 2010 ) 012034 f. liao , y .- k .
jan , jour . of rehab .
res . and develop . 48 ( 2011 ) 787 g. r. jafari , p. pedram , l. hedayatifar , j. stat .
( 2007 ) p04012 p. owicimka , j. kwapie , i. celiska , s. drod , r. rak , arxiv:1106.2902v1 [ physics.data-an ] f. a. hirpa , m. gebremichael , t. m. over , water resour .
46 ( 2010 ) w12529 k. matia , y. ashkenazy , and h. e. stanley , europhys .
61 ( 2003 ) 422 p. owicimka , j. kwapie , and s. drod , physica a 347 ( 2005 ) 626 j. kwapie , p. owicimka , and s. drod , physica a 350 ( 2005 ) 466 j. jiang , k. ma , and x. cai , physica a 378 ( 2007 ) 399 k. e. lee and j. w. lee , physica a 383 ( 2007 ) 65 g. lim , s. kim , h. lee , k. kim , and d .- i .
lee , physica a 386 ( 2007 ) 259 .
czarnecki , d. grech , act .
phys . pol .
a 117 ( 2010 ) 4 j. w. kantelhardt , arxiv:0804.0747v1 [ physics.data-an ] ( 2008 ) j. ludescher , m. i. bogachev , j. w. kantelhardt , a. y. schumann , a. bunde , physica a 390 ( 2011 ) 2480 .
d. gulich , l. zunino , physica a 391 ( 2012 ) 4100 j. feder , fractals , new york , plenum press ( 1988 ) .
peitgen , h. jrgens , d. saupe , chaos and fractals , @xmath120nd ed .
springer ( 2004 ) .
d. grech and g. pamua , acta phys . pol .
a 121 ( 2012 ) b-34 d. grech , g. pamua , physica a 392 ( 2013 ) ( in print ) c. meneveau , k. r. sreenivasan , phys .
59 , ( 1987 ) 1424 a .-
barabasi , t. vicsek , multifractality of self - affine fractals , phys .
a 44 , ( 1991 ) 2730 .
d. grech , g. pamua , arxiv:1307.2014 [ physics.data-an ] | we provide an alternative method for analysis of multifractal properties of time series .
the new approach takes into account the behavior of the whole multifractal profile of the generalized hurst exponent @xmath0 for all moment orders @xmath1 , not limited only to the edge values of @xmath0 describing in mfdfa scaling properties of smallest and largest fluctuations in signal .
the meaning of this new measure is clarified and its properties are investigated for synthetic multifractal data as well as for real signals from stock markets . we show that the proposed new measure is free of problems one can meet in real nonstationary signals , while searching their multifractal signatures .
@xmath2 * keywords * : multifractality , spurious multifractality , finite size effects , multifractal detrended analysis , multifractal bias , scaling , time series analysis , new multifractal measure , generalized hurst exponent + * pacs : * 05.45.tp , 89.75.da , 05.40.-a , 89.65.gh,89.75.-k + |
the star formation ( sf ) is a fundamental process in the evolution of galaxies and is far from being well understood .
the sf is usually characterized by the initial mass function ( imf ) and the total sf rate ( sfr ) , which depends on many factors such as the density of the interstellar gas , its morphology , its metallicity , _ etc . _
generally , four major factors drive star formation in galaxies : large scale gravitational instabilities , cloud compression by density waves , compression in a rotating galactic disk due to shear forces , and random gas cloud collisions . in galaxies with previous stellar generations
additional sf triggers exist , such as shock waves from stellar winds and supernova explosions . in dense environments , such as clusters of galaxies and compact groups , tidal interactions , collisions with other galaxies ,
ism stripping , and cooling flow accretion probably play some role in triggering the sf process .
the triggering mechanisms were reviewed recently by elmegreen ( 1998 ) .
while `` global '' phenomena play a large part in grand design spirals , random collisions of interstellar clouds have been proposed as one explanation for dwarf galaxies with bursts of sf . due to their small size , lack of strong spiral pattern , and sometimes solid - body rotation ( _ e.g. , _ martimbeau 1994 , blok & mcgaugh 1997 ) , the star formation in dwarf galaxies
can not be triggered by compression from gravitational density waves or by disk shear .
therefore , understanding sf in dwarf galaxies should be simpler than in other types of galaxies , because the number of possible trigger mechanisms is reduced .
the h@xmath0 emission from a galaxy measures its ongoing sfr ( kennicutt 1983 , kennicutt 1994 ) .
the blue luminosity of a galaxy measures its sf integrated over the last @xmath1 yrs ( gallagher 1984 ) .
the red continuum radiation originates both from relatively young stars which already evolved into red giants and super - giants , and from a large population of aged low - mass stars , if previous sf episodes took place . hunter ( 1998 )
tested a set of sf predictors on two small samples of dwarf galaxies , one observed by them and another derived from de blok ( 1997 ) .
they found that the ratio of hi surface density to the critical density for the appearance of ring instabilities did not correlate with the star formation , but that the stellar surface brightness did . from this , they concluded that possibly some stellar energy input provides the feedback mechanism for star formation .
brosch ( 1998 ) confirmed that the strongest correlation among a number of parameters tested on a sample of virgo cluster dwarf irregular galaxies was between the average h@xmath0 surface brightness and the mean blue surface brightness .
this is similar to the findings of phillipps & disney ( 1985 ) for spiral galaxies , where in a sample of 77 spiral galaxies from kennicutt & kent ( 1983 ) a correlation was found between the total h@xmath0 emission ( expressed as specific sfr or as h@xmath0 equivalent width ) and the average blue surface brightness . on the level of individual hii regions in dwarf
irregular galaxies , heller ( 1998 : hab98 ) showed that a correlation exists also between the h@xmath0 line flux and the red continuum flux underneath the region , measured with the same aperture as the line flux .
we emphasize that in this case the correlation is between local quantities , not for overall galactic properties .
these correlations indicate that the regulation of sf in irr s is local and by the existing stellar population
. the self - regulated evolution of dwarf galaxies has recently been modeled by andersen & burkert ( 1997 ) .
we concentrate here on samples of late - type dwarf irregular galaxies ( digs ) in the virgo cluster ( vc ) and elsewhere in the nearby ( within 100 mpc ) universe .
the reason for selecting digs is to limit the number of possible sf trigger mechanisms ; digs are devoid of large - scale sf triggers , as explained above .
we take advantage of the availability of net h@xmath0 line images to determine the general pattern of the distribution of hii regions over the irregular galaxies . as far as we could ascertain , such a study of a large sample of irregular galaxies was never published .
previous attempts to classify digs were _
e.g. , _ by sandage & binggeli ( 1984 ) for low surface brightness ( lsb ) objects , by loose & thuan ( 1985 ) for bcds , and by patterson & thuan ( 1996 ) for digs .
all used broad - band images to perform the classification .
we do not study here the morphology of the galaxies , as reflected by their light distribution on broad - band images , but rather the morphology of their ensemble of hii regions .
our goal is to gain some insight on the star - forming properties of this class of galaxies .
the question of the spatial distribution of hii regions in irregular galaxies has been studied previously by hodge ( 1969 ) for seven nearby objects , and by hunter & gallagher ( 1986 ) for a larger sample of galaxies .
the primary sample consists of 52 digs in the vc with hi measurements from hoffman _
( 1987 , 1989 ) . the sample was constructed in order to enable the detection of weak dependencies of the star formation properties on the hydrogen content and on the surface brightness .
we selected two sub - samples by surface brightness ; one represents a high surface brightness ( hsb ) group and is classified as either bcd or anything+bcd , and another represents a low surface brightness ( lsb ) sample and includes only imiv or imv galaxies . the uniform morphological classification , which bins the digs in the hsb or lsb groups , is exclusively from binggeli ( 1985 , vcc ) .
the galaxies were observed at the wise observatory ( wo ) from 1990 to 1997 , with ccd imaging through broad bands and narrow h@xmath0 bandpasses in the rest frame of each galaxy .
the discussion of all observations and their interpretation is the subject of other papers ( ab98 , hab98 ) .
we restrict the discussion here to the localization of the star - forming regions on the broad - band , or continuum light images of the digs . in some virgo galaxies we did not detect h@xmath0 emission ; these objects have been omitted in table 1 leaving 30 digs from our combined virgo sample .
the two virgo cluster samples are augmented here by 83 irregular galaxies for which data were collected from the literature .
not all these objects are dwarfs but all appear to be , or are classified as , irregular galaxies ; we call them here digs and are not strict in qualifying an object as `` dwarf ''
. three objects of the extended sample appear in two references ; these objects have been classified independently and have two entries in table 1 .
the total number of classifications is thus 124 , but only 110 different objects have been considered .
we inspected h@xmath0 and broad - band or red continuum images from strobel ( 1991 ) , miller & hodge ( 1994 ) , mcgaugh ( 1995 ) , van zee ( 1996 ) , marlowe ( 1997 ) , martin ( 1997 ) , hilker ( 1998 ) , and gavazzi ( 1998 ) .
the images from van zee include objects analyzed in her phd thesis ( van zee 1996 ) and some galaxies from an unpublished comparison sample .
the objects were selected to be lsb digs and were checked not to show obvious signs of interaction on the palomar sky survey plates .
the objects from van zee contribute 27 galaxies to the extended sample .
the objects studied by gavazzi ( 1998 ) and included here consist of eight galaxies classified as irr , most in the a1367 cluster . the hilker ( 1997 ) object is an lmc - like galaxy in the fornax cluster . ho ii was studied in detail by puche ( 1992 ) ; we used the published images for the present classification .
the 12 galaxies studied by marlowe ( 1997 ) are classified as either amorphous or blue compact , are intrinsically faint ( _ i.e. , _ dwarf ) , and are nearby . the sample of martin ( 1997 ) contains a heterogeneous assemblage of star - forming dwarf galaxies .
three objects with spiral morphology ( n2537 , vii zw403 , and n4861 ) from her list were excluded from the present analysis , leaving 12 galaxies to be considered here .
additionally , four objects were added from the study of dwarf galaxies in the m81 system ( miller & hodge 1994 ) and four other from the study of digs by strobel ( 1991 ) .
finally , we included 13 objects from the morphological study of lsb disk galaxies of mcgaugh ( 1995 ) which did not show strong spiral patterns on the published images .
the only restrictions to the inclusion of a galaxy in the extended sample were that the object should be classified as an irregular galaxy in the original publication and that it would have a net h@xmath0 and an off - line image .
this resulted in a very heterogeneos collection of irregular galaxies ; those from martin ( 1997 ) and marlowe ( 1997 ) are mainly low - luminosity , nearby objects , while those from gavazzi ( 1998 ) , being at @xmath270 mpc , are of high luminosity and would not be strictly classifiable as `` dwarfs '' .
most of the galaxies in gavazzi were rejected because they were not irr galaxies and so were all the objects analyzed by hunter ( 1998 ) , which do not have published images but only azimuthal averages of line and continuum emission .
we could not use the large h@xmath0 and continuum image set in koopman ( 1997 ) because it contains only spiral and lenticular galaxies .
the entire selection of 110 galaxies classified here is listed in table 1 , where the objects from our virgo sample are identified by their number in the virgo cluster catalog ( binggeli 1985 ) .
the analysis reported here is based primarily on the visual inspection of the net h@xmath0-line and off - line images of each galaxy .
these are usually presented side - by - side in the original publications , at the same scale and with enough `` gray - scale stretch '' to allow easy perception of the hii regions on the net - line image , and of the general outline of the galaxy in the off - line image .
this facilitates the comparison and the determination of whether the hii regions are distributed mostly at the edges or near the center of an object .
these two cases have been noted in table 1 as e for edge and c for center , and are the primary morphological index used for this classification .
there are a few cases of mixed morphology , which have been so noted in table 1 .
one such example is ugc 7178 , from the primary sample of van zee ( 1996 ) , which has four hii regions , two near the center and two at the edge and is classified here as c+e .
other galaxies do not show preferential sf at either the center or the edge , but display a @xmath2linear distribution of hii regions showing up as a `` spine '' on the galaxy image .
these objects are noted as l = linear , and/or sp = spine types and are probably related to the `` cometary '' galaxies noted by loose & thuan ( 1985 ) .
a few objects have a number of hii regions arranged on the ( partial ) circumference of an ellipse ; these are marked el = ellipse in table 1 .
finally , some irr s show a scattering of hii regions and are accordingly marked d = diffuse .
ugc11820 in van zee s primary sample ( van zee 1996 ) has its hii regions arranged on @xmath2a spiral arm .
it is possible that this is a case of mistaken classification and the object is probably a spiral , as listed in ugc and in ned .
a similar case may be object 127037 in gavazzi ( 1998 ) . our classification , by the distribution of the visible hii regions ,
should be compared with that of patterson & thuan ( 1996 ) , where six classes of digs were distinguished on the basis of broad - band b and i images .
the classifications are an extension of the loose & thuan ( 1986 ) scheme and bin the digs into dwarf spirals ( ds ) , nucleated dwarf irregulars ( dine ) , dwarf ellipticals with a central ridge ( dire ) , dwarfs with `` asymmetric star formation '' ( dia ) , objects with randomly scattered star formation similar to the gr8 galaxy ( gr8 ) , and dwarfs which show a bar with whispy extensions ( dib ) .
the two ugc objects in common , u300 and u2162 , have both been classified by patterson & thuan as dia , while we classify them as e and a , confirming the asymmetry mention and adding the qualifier that the hii regions tend to be at the edges of the galaxies .
we used a second morphological index to flag a symmetric or asymmetric distribution ( s / a ) of hii regions .
a galaxy is labelled asymmetric ( a ) in the distribution of its hii regions if these are located predominantly on one side of the galaxy . in other words ,
the label a is assigned if it is possible to draw on the continuum or broad - band image of a galaxy a diameter which bisects it so that most of the hii regions are on one side of this diameter . if no such diameter seems to exist , the galaxy is classified as symmetrical ( s ) in the distribution of its hii regions .
this asymmetry criterion is similar to that used by hodge ( 1969 ) , with the exception that hodge used the `` reference frame '' of the hii region distribution while we used that of the red continuum light .
note that the secondary classification of the asymmetry is independent of the primary classification of edge / center / spine described above .
this secondary classifying index is also listed in table 1 . in some cases the images were too poor to allow a classification .
these objects have a question mark in the table .
the references for the images are listed in table 2 .
we show in figure 1 examples of the primary and secondary classifications using objects from ab98 and hab98 .
the top row shows an object with edge distribution of hii regions , which is asymmetric .
the middle row shows a galaxy where the hii regions are arranged in a linear , spine - like configuration .
the bottom row shows an object with a centrally located , symmetric distribution of hii regions .
obviously , a consistent classification requires similar types of display , contrast of images , _ etc .
_ this is not always the case , as some sources did not provide two images ( line and continuum ) for a galaxy , but only one with overlaid contours for the missing information ( _ e.g. , _ martin 1997 or marlowe 1997 ) and we also did not have control over the display mode .
nevertheless , the few galaxies appearing in two references allow some measure of confidence in the classification : n1800 , n5253 , and ii zw40 are common to the samples from marlowe ( 1997 ) and martin ( 1997 ) ; their classification , performed completely independently on different images , is virtually identical for all three cases . to put the classification on a more `` objective '' and numerical basis , and to aviod possible biases caused by the tendency to detect structures when visually inspecting random distributions of dots , we formed two indices which quantify the degree of concentration of hii regions ( `` concentration index''=ci ) and the amount of asymmetry in the hii region distribution ( `` asymmetry index''=ai ) .
these indices are calculated from counts of hii identified in various regions of the galaxies net - line images we inspected and are listed in table 1 .
the definition of these two parameters is fairly intuitive and is explained below .
ci is defined as the ratio of the number of hii regions in the inner part of the galaxy to that in the outer part .
we count the hii regions within half the semi - major axis from the center , and divide this by one - quarter of the number of hii regions external to this region .
the one - quarter factor brings the comparison to a number per equal - area basis and is , in fact , a ratio of the number surface density of hii regions .
ci can have values between zero and infinity , as the outer part of the galaxy may be devoid of hii regions .
a value of unity represents a uniform distribution of hii regions , while galaxies with no emission near the center have ci=0 .
galaxies with emission localized in their centers and no emission detected in their outer parts have ci=@xmath3 ; this is represented in the table as ci=100 .
the asymmetry index ai is defined as the ratio of the number of hii regions counted in the `` hii poor '' half of the galaxy to that in the `` hii rich '' area , where the divider is the bisecting diameter selected visually to show the largest contrast between the two galaxy halfs .
ai ranges from 0 to 1 , with unity representing a symmetric distribution of hii regions .
the smaller the value of ai , the more asymmetric is the distribution of hii regions .
we explain the two - dimensional morphological classification with the example of vcc 17 , shown in the top row of images in fig .
vcc 17 has two central hii regions and four regions in its outer part .
its ci is therefore @xmath4 .
a diameter may be drawn on the figure which puts two hii regions on one side of it and four on the other side .
this appears to be the most extreme asymmetry , making ai=@xmath5=0.5 .
our choice of irregular galaxies in which to study the patterns of star formation was quite deliberate .
as mentioned in the introduction , these objects should be devoid of large - scale sf triggering mechanisms such as density waves or various disk instabilities .
therefore the triggering mechanisms should be simpler to disentangle in this kind of objects .
the star formation indicator used here was the h@xmath0 line emission and the distribution of the hii regions was checked against the light distribution of either the red continuum , or any broad - band images supplied by the authors of a specific paper .
the summary statistic for the distribution among types and the symmetric or asymmetric morphologies is presented in table 3 .
the galaxies with mixed morphologies have been counted once in each bin , thus the total number of cases listed in the table is larger than the number of actual galaxies inspected .
galaxies with incomplete or dubious classification , which have only a question mark in the s / a column of table 1 , have been excluded from the statistic .
table 3 shows that most galaxies have an e - type distribution of hii regions .
specifically , we find that about half of all classifications are e and a. the only other bin populated by a significant amount of irregular galaxies is c and s ( @xmath2one quarter of all classifications ) and the other bins are essentially empty . most galaxies , which were selected only to be irregulars based on their appearance on broad - band images , form preferrentially their hii regions in their outer regions ( type a has @xmath22/3 of the cases ) .
the objects in which the distribution of hii regions appears symmetrical are mostly those where it is also central .
this is very similar to what loose & thuan ( 1985 ) found for bcds .
we show the distribution of the two morphological indices in figures 2 and 3 ( the latter shown to emphasize the behavior of the morphological indices for ci@xmath60 ) .
it is clear that there is a strong dychotomy , because of the objects with essentially central h@xmath0 emission which make up the rightmost part of the plot ( ai@xmath7 ) .
the distribution of the objects with inner * and * outer hii regions is shown in figure 4 .
the figure shows the dominance of the highly asymmetric coverage of digs by hii regions , with most cases concentrating at [email protected] .
this implies that in most digs one half of the galaxy has twice or more the number of hii regions than the other half .
gerola & seiden ( 1978 ) proposed that the mechanism regulating the sf in irr s is the stochastic self - propagating sf ( sspsf ) .
their simulations , as well as the more recent ones by jungwiert & palous ( 1994 ) , produce preferentially flocculent or grand - design spirals .
if such a mechanism operates in digs it should produce an expanding sf wave which would engulf the entire galaxy , or at least those regions with suitable ism density .
unfortunately , the 3d simulations of sspsf relevant to digs ( comins 1983 , 1984 ) do not show `` snapshots '' of the sf proceeding with time through the galaxy .
such plots could have been analyzed in a similar manner to the galaxy images and possibly some constraint could have been derived .
the few papers which do show such plots ( _ e.g. , _ gerola 1980 ) have too few figures to make their analysis statistically significant . note also that in large , slowly - rotating disks , like the lmc , the sspsf tends to produce mostly long filaments of young stars which show no special preference of galactic location ( feitzinger 1987 ) .
if the galaxy is small , and a number of sns explode off its center , it is possible in principle to have a compression wave travel through the gas and form stars in suitable location while escaping from the galaxy in places where the ism is thin or altogether absent .
it is difficult to estimate the likelihood of such a mechanism but it is possible to examine it in a well - resolved object .
the nearby dig ho ii was studied intensively by puche ( 1992 ) .
a comparison of their h@xmath0 and off - line images shows that ho ii forms stars near its center .
the hi synthesis map indicates that the h@xmath0 emission originates either at the interfaces between large holes in the hi distribution or in the small hi holes .
thus this case also does not argue in favour of the sspsf forming stars asymmetrically , or at the edges of a galaxy .
if the ism in an irregular galaxy is preferentially aligned with its long dimension , a spine of hii regions could form by the sspsf mechanism . similarly ,
if the galaxy is in a symmetrical gravitational potential well , its ism could concentrate at the bottom of the potential and begin forming stars there in a c configuration .
is there a mechanism of sf which would produce predominantly lop - sided regions of sf at the edge of a galaxy ?
the question was posed recently for the lmc by de boer ( 1998 ) , who proposed a mechanism to explain the observed distribution of giant sf structures lining the edge of the lmc .
they postulate that the interaction between the lmc gas and gas in the halo of the milky way causes gas compression followed by star formation .
the sf takes place at the interface between the two gas distributions , where the interaction beweeen the lmc gas and the mw gas occurs .
the rotation of the lmc moves the regions with newly formed stars away from the place of formation , the star formation ceases , and the newly formed stars simply age .
we tested this possibility with the objects in our virgo samples which have measurements of h@xmath0 and underlying red continuum emission for individual hii regions .
the test is reported in detail in hab98 .
if the mechanism of de boer ( 1998 ) is at work , we expect a decrease of the h@xmath0 line intensity simultaneously with an enhancement of the red stellar continuum under the hii region as it ages .
the test we performed involved a comparison of the ratios of h@xmath0 intensities and of the red continua for the brightest and the faintest hii regions identified in the same galaxy .
these ratios were compared for 13 objects with multiple hii regions and a trend in the opposite direction from that expected was found ; the more intense the underlying red continuum , the stronger the line emission is .
this argues against the de boer ( 1998 ) proposition and in favour of a mechanism regulating the star formation through the existing local stellar population .
mcgaugh ( 1995 ) mention that the m=1 density wave mode may be possible in lsb galaxies , giving rise to one - armed spirals . in principle , it is possible that the excess of irregulars with asymmetric , edge - concentrated star formation is due to this phenomenon , but we deem it unlikely .
the reason is that if the sf process would be driven by a density wave , the one - armed spiral pattern should be visible not only in the distribution of hii regions but also in the red continuum image .
a possibly related phenomenon , of displaced light centers with respect to the outermost isophotes in a sample of extremely late - type spirals was recently reported by matthews & gallagher ( 1997 ) .
the phenomenon was explained as a consequence of the disk , which is the luminous galaxy , orbiting in an off - center position within an extended dark halo ( levine & sparke 1998 ) .
the question of dark matter ( dm ) halos in the context of irregular galaxies has also been discussed by hunter ( 1998 ) , where the conclusion was that the dm may affect the star formation by enhancing the gravitational potential .
it is possible that such a model could explain also the asymmetric patterns of star formation which we reported above , but its exploration , with special emphasis on local density enhancements by the dm , is beyond the scope of this paper .
we analysed the distribution of regions where star formation takes place at present in a sample of 110 irregular galaxies .
our results can be summarized as follows : 1 .
star formation takes place predominantly at the edges of dwarf irregular galaxies , mostly to one side of a galaxy .
existing models of star formation in such objects do not predict such a distribution of star forming regions over a galaxy .
the proposals of de boer ( 1998 ) , of an interaction between the galaxy and some surrounding medium which compresses the ism and thus enhances the star formation , and of mcgaugh ( 1995 ) , of strong single - arm spiral patterns in dwarf galaxies which could give rise to asymmetric distribution of star - forming regions , can probably be rejected .
no good explanation was identified for the peculiar location of the hii regions in irregular galaxies .
nb is grateful for continued support of the austrian friends of tel aviv university and for the hospitality of the space telescope science institute , where most of this paper was written .
ea is supported by a special grant from the ministry of science to develop tauvex , a uv imaging experiment .
ah acknowledges support from the us - israel binational science foundation and travel grants from the sackler institute for astronomy . astronomical research at tel aviv university is partly supported by a grant from the israel science foundation .
discussions on this subject with mario livio appreciated .
we are grateful for constructive remarks on this subject from lyle hoffman , crystal martin , and an anonymous referee .
liese van zee and giuseppe gavazzi kindly supplied electronic copies of images from their samples of galaxies , used for some of the comparisons presented here .
almoznino , e. & brosch , n. 1998 , mnras , in press ( ab98 ) .
andersen , r .-
p . & burkert , a. 1997 , apj , preprint .
brosch , n. , heller , a. & almoznino , e. 1998 , apj , 504 ( september 10 ) , in press .
binggeli , b. , sandage , a. & tamman , g.a .
1985 , aj , 90 , 1681 .
comins , n.f .
1983 , apj , 266 , 543 .
comins , n.f .
1984 , apj , 284 , 90 .
de blok , e. & mcgaugh , s.s .
1997 , mnras 290 , 533 .
de boer , k.s . , braun , j.m . ,
vallenari , a. & mebold , u. 1998 , a&a , 329 , l49 .
gerola , h. & seiden , p.e .
1978 , apj , 223 , 129 .
gerola , h. , seiden , p.e . & schulman , l.s .
1980 , apj , 242 , 517 .
gallagher , j.s . ,
hunter , d.a .
& tutukov , a.v .
1984 , apj , 284 , 544 .
gavazzi , g. , catinella , b. , carraso , l. boselli , a. & contursi , a. 1998 , aj , 115 , 1745 .
elmegreen , b.g .
1998 , in _ origins of galaxies , stars , planets and life _
woodward , h.a .
thronson , & m. shull , eds . ) , asp series , in press .
feitzinger , j.v . ,
haynes , r.f . , klein , u. , wielebinski , r. & perschke , m. 1987 , vistas in astr .
, 30 , 243 .
heller , a. , almoznino , e. & brosch , n. 1998 , mnras , submitted ( hab98 ) .
hilker , m. , bomans , d.j . ,
infante , l. & kissler - patig , m. 1997 , a&a , 327 , 562 .
hodge , p. 1969
, apj , 156 , 847 .
hoffman , g.l .
, helou , g. , salpeter , e.e . ,
glosson , j. & sandage , a. 1987 , apjs , 63 , 247 .
hoffman , g.l . ,
lewis , b.m . ,
helou , g. , salpeter , e.e . &
williams , h.l .
1989 , apjs , 69 , 65 . hunter , d.a . & gallagher , j.s .
1986 , pasp , 98 , 5 .
hunter , d.a . ,
elmegreen , b.g . & baker , a.l .
1998 , apj , 493 , 595 .
jungwiert , b. & palous , j. 1994 , a&a , 287 , 55 .
kennicutt , r.c .
1998 , apj , 498 , 541 .
kennicutt , r.c .
1983 , apj , 272 , 54 .
kennicutt , r.c . ,
tamblyn , p. & congdon , c.w .
1994 , apj , 435 , 22 .
levine , s.e .
& sparke , l.s .
1998 , apjl , in press ( astro - ph/9803146 ) .
loose , h .- h . & thuan , t.x .
1985 in _ star - forming dwarf galaxies _
( d. kunth , t.x .
thuan & j. tran than van , eds . ) , gif sur yvette : editions frontieres , p. 73 .
marlowe , a.t .
, meurer , g.r .
& heckman , t.m .
1997 , apjs , 112 , 285 .
martin , c. 1997 , apj , 491 , 561 .
martimbeau , n. , carignian , c. & ray , j .-
1994 , aj , 107 , 543 .
matthews , l.d . & gallagher , j.s .
1997 , aj , 114 , 1899 .
mcgaugh , s.s . , schombert , j.m . & bothun , g.d .
1995 , aj , 109 , 2019 .
miller , b.w .
& hodge , p. 1994 , apj , 427 , 656 .
patterson , r.j .
& thuan , t.x .
1996 , apjs , 107 , 103 .
phillipps , s. & disney , m. 1985 , mnras , 217 , 435 .
puche , d. , westphal , d. , brinks , e. & roy , j .- r .
1992 , aj , 103 , 1841 .
schmidt , m. 1959 , apj , 129 , 243 .
strobel , n.v . ,
hodge , p. & kennicutt , r.c .
1991 , apj , 383 , 148 .
tresse , l. & maddox , s.j .
1998 , apj , 495 , 691 .
van zee , l. 1996 , phd thesis , cornell university .
* figure 1 : examples of galaxies of different types and morphologies .
each row shows the galaxy as imaged through the h@xmath0 filter at left , through the continuum filter at the center , and in the net - h@xmath0 line at right .
the galaxies are vcc 17 ( e & a ) , vcc 1374 ( l , sp & s ) , and vcc 10 ( c & s ) . a 10 arcsec bar at the upper right corner of each row sets the scale .
* figure 2 : distribution of classification indices for digs .
each independent classification is represented by a circle .
the points in regions of high concentration have been `` jiggled '' slightly in an attempt to separate points which are very close together .
concentration index values of 100 represent galaxies with ci=@xmath3 ( _ i.e. , _ all hii regions in central locations ) .
* figure 3 : distribution of classification indices for digs .
this is an expansion of figure 2 near ci@xmath60 .
the points have * not * been jiggled in this presentation .
* figure 4 : distribution of the asymmetry index among galaxies which have ci@xmath9 .
most of the objects concentrate in the low ai bins , indicating a preference for asymmetric distribution of hii regions .
u300 & e & 0 & a & 0.5 & 1 & u191 & e & 0 & a & 0.58&2 u521 & c & 100 & s & 1 & 1 & u634 & e & 0 & a & 0 & 2 u2684 & e+c & 100 & a & 1 & 1 & u891 & l , sp & 100 & s&1&2 u2984 & e , el & 0 & a & 0.38 & 1 & u1175 & e&0 & a & 0&2 u3174 & e & 1.33 & a & 0.4 & 1 & u2162 & e & 0 & a&0.5 & 2 u3672 & e & 0 & a & 0 & 1 & u3050 & c+e&1.33 & a&0.5&2 u4660 & d & 0 & a & 0.33 & 1 & u4762 & e & 0&a & 0.5&2 u5716 & e , el & 0&a & 0 & 1 & u5764 & c+e & 0&a & 0&2 u7178 & c+e & 0.67 & a&0.17 & 1 & u5829 & e & 0&a & 0&2 ugca357 & e & 0 & a&0 & 1 & u7300 & e & 0&a & 0.33 & 2 haro43 & l , sp & 2 & s&1 & 1 & u8024 & e & 0 & s?&0.29&2 u9762 & d ? & - & ? & - & 1 & u9128 & e & 0 & s&0 & 2 u10281 & e & 0&a & 0 & 1 & ddo210 & c & 100&s&0 & 2 u11820 & e & 0.87 & a&0.58 & 1 & n1427a & e & 0.2 & a&0.11&4 97073 & e , el & 0.31 & a & 0 & 3 & v0017 & e & 2&a & 0.5&7 97079 & e & 0 & a&0 & 3 & v0169 & e & 0&a & 0.5 &
7 97087 & l , sp & 1&a&1 & 3 & v0217 & e&0 & a & 0.5&7 97138 & e&0.66 & a&0.25 & 3&v0328 & e & 0&a & 0.5 & 7 108085 & e & 0&a & 0 & 3&v0350 & c ?
& 0 & s & 0&7 127037 & e , el&0.57 & a&0.20 & 3&v0477 & e & 0&a & 0&7 160086 & c & 0.57&s & 0.75 & 3 & v0530 & c & 100&a&0&7 160139 & e & 0&a & 0.22 & 3&v0826 & e&0 & a & 0.5&7 haro 14 & c & 100&a&1 & 5 & v0963 & c & 100&s&0&7 n625 & c & 100&s & 1 & 5 & v1455 & c & 100 & a&0 & 7 n1510 & c & 100&s&1 & 5&v1465 & c+e & 2.5 & a & 0&7 n1705 & c & 100&s&1 & 5&v1468 & e & 0&a & 1&7 n1800 & c+e & 1 & s&0 & 5 & v1585 & e & 0&a & 0&7 n2101 & c+e & 0.33&s & 0.23 & 5 & v1753 & e & 0&a & 0&7 ii zw40 & c & 1 & s & 1 & 5&v1952 & e & 0&a&0 & 7 n2915 & c+e & 100&a & 0.66 & 5&v1992 & e & 0&a & 0&7 n3125 & c+e & 1&a & 1 & 5&v2034 & d ? & - & ? & - & 7 n3955 & c & 100&a?&1 & 5 & v0010 & c&100&s & 1 & 8 n4670 & c & 100&s & 1 & 5 & v0144 & c&100&s&1 & 8 n5253 & c & 100&s & 1 & 5&v0172 & c+e&0&a&0 & 8 n1569 & c & 1&s&0.33 & 6&v0324 & c&100&s&1 & 8 n1800 & e & 100&a & 1 & 6 & v0410 & c&100&s&1 & 8 ii zw 40 & c & 1&s & 0.25 & 6 & v0459 & e&0&a & 1&8 n2363 & c & 100&s & 1 & 6 & v0513 & c&100&s&1 & 8 i zw 18 & c & 4&s & 0 & 6&v0562 & c & 100 & s & 1 & 8 m82 & c & 100&s & 1 & 6 & v0985 & e&100&a&1 & 8 n3077 & c & 100&s & 1 & 6 & v1179 & c&100&s&1 & 8 sex a & e & ? & a & ? & 6&v1374 &
l , sp & 2&a&1 & 8 n3738 & c & 100&a & 1 & 6&v1725 & l , sp & 4&a&0 & 8 n4214 & l , sp & 0.5&s ? & 0.33 & 6&v1791 & e&0&a & 0.5&8 n4449 &
l , sp & ? & a & ? & 6&ho ii & c&1 & s&1 & 9 n5253 & c & 100 & s & 1 & 6&ddo 47 & d&1.28 & s & 0.89&10 ho i & e & 0.13&a & 0.35 & 11 & sex b & e & 0.36 & a & 0.5&10 ho ix & e & 0&a & 0.20 & 11 & ddo 167 & c & 25&a & 0.66 & 10 ic 2547 & e & 0&a & @xmath20.5&11 & ddo
168 & l , sp & 1 & s & 0.71&10m81db & l , sp & 0.25&s & 0.33 & 11 & ddo 187 & e & 0 & a & 0.5&10f415 - 3 & c & 2 & a & 0 & 12 & f469 - 2 & c+e & 2 & a & 0 & 12 f561 - 1 & e & 1.14 & a & 0.75 & 12 & f563-v1 & d & 100 & s & 1 & 12 f611 - 1 & e & 0 & a & 0 & 12 & u12695 & e & 0.5 & a & 0.14 & 12 f562-v2 & c+e & 0 & a & 0.5 & 12 & f746 - 1 & e & 0.36 & a & 0.87 & 12 u5709 & c & 100 & s & 1 & 12 & f568 - 1 & c+e & 0.28 & s & 1 & 12 f583 - 5 &
l , sp & 0 & s & 0.75 & 12 & f585 - 3 & c & 0.8 & s & 0.66 & 12 u5675 & c & 0 & s & 1 & 12 1 & van zee 1996 primary & 142 & van zee 1996 secondary & 13 3 & gavazzi 1998 & 84 & hilker 1998 & 1 5 & marlowe 1997 & 12 6 & martin 1997 & 12 7 & heller 1998 & 17 8 & almoznino & brosch 1998 & 13 9 & puche 1992 & 1 10 & strobel 1991 & 4 11 & miller & hodge 1994 & 4 12 & mcgaugh 1995 & 13 & total number of images & 113 | the location of hii regions , which indicates the locus of present star formation in galaxies , is analyzed for a large collection of 110 irregular galaxies ( irr ) imaged in h@xmath0 and nearby continuum .
the analysis is primarily by visual inspection , although a two - dimensional quantitative measure is also employed .
the two different analyses yield essentially identical results .
hii regions appear preferentially at the edges of the light distribution , predominantly on one side of the galaxy , contrary to what is expected from stochastic self - propagating star formation scenarios .
this peculiar distribution of star forming regions can not be explained by a scenario of star formation triggered by an interaction with extragalactic gas , or by a strong one - armed spiral pattern .
# 1#2#3#4#5#6#7 to#2 ' '' '' |
eit is a non - invasive real - time functional imaging modality for the continuous monitoring of physiological functions such as lung ventilation and perfusion .
the image contrast represents the time change of the electrical conductivity distribution inside the human body . using an array of surface electrodes around a chosen imaging slice ,
the imaging device probes the internal conductivity distribution by injecting electrical currents at tens or hundreds of khz .
the injected currents ( at safe levels ) produce distributions of electric potentials that are non - invasively measured from the attached surface electrodes .
a portable eit system can provide functional images with an excellent temporal resolution of tens of frames per second .
eit was introduced in the late 1970s@xcite , likely motivated by the success of x - ray ct .
numerous image reconstruction methods and experimental validations have demonstrated its feasibility@xcite and clinical trials have begun especially in lung ventilation imaging and pulmonary function testing@xcite .
however , eit images often suffer from measurement noise and artifacts especially in clinical environments and there still exist needs for new image reconstruction algorithms to achieve both high image quality and robustness . the volume conduction or lead field theory indicates that a local perturbation of the internal conductivity distribution alters the measured current voltage or trans - impedance data , which provide core information for conductivity image reconstructions . the sensitivity of the data to conductivity changes varies significantly depending on the distance between the electrodes and the conductivity perturbation : the measured data are highly sensitive to conductivity changes near the current - injection electrodes , whereas the sensitivity drops rapidly as the distance increases@xcite
in addition , the boundary geometry and the electrode configuration also significantly affect the measured boundary data .
the sensitivity map between the measured data and the internal conductivity perturbation is the basis of the conductivity image reconstruction .
a discretization of the imaging domain results in a sensitivity or jacobian matrix , which is inverted to produce a conductivity image .
the major difficulty arises from this inversion process , because the matrix is severely ill - conditioned .
conductivity image reconstruction in eit is , therefore , known to be a fundamentally ill - posed inverse problem .
uncertainties in the body shape , body movements , and electrode positions are unavoidable in practice , and result in significant amounts of forward and inverse modeling errors .
measurement noise and these modeling errors , therefore , may deteriorate the quality of reconstructed images .
numerous image reconstruction algorithms have been developed to tackle this ill - posed inverse problem with data uncertainties@xcite . in the early 1980s , an eit version of the x - ray ct backprojection algorithm@xcite
was developed based on a careful understanding of the ct idea .
the one - step gauss - newton method@xcite is one of the widely used methods and often called the linearized sensitivity method .
several direct methods were also developed : the layer stripping method@xcite recovered the conductivity distribution layer by layer , the d - bar method@xcite was motivated from the constructive uniqueness proof for the inverse conductivity problem@xcite , and the factorization method@xcite was originated as a shape reconstruction method in the inverse scattering problem@xcite .
recently , the discrete cosine transform was adopted to reduce the number of unknowns of the inverse conductivity problem@xcite .
there is an open source software package , called eidors , for forward and inverse modelings of eit@xcite .
there are also novel theoretical results showing a unique identification of the conductivity distribution under the ideal model of eit@xcite .
most common eit image reconstruction methods are based on some form of least - squares inversion , minimizing the difference between the measured data and computed data provided by a forward model .
various regularization methods are adopted for the stable inversion of the ill - conditioned jacobian matrix .
these regularized least - squares data - fitting approaches adjust the degree of regularization by using a parameter , controlling the trade - off between data fidelity and stability of reconstruction .
their performances , therefore , depend on the choice of the parameter . in this paper , we propose a new regularization method , which is designed to achieve satisfactory performances in terms of both fidelity and stability regardless of the choice of the regularization parameter . investigating the correlations among the column vectors of the jacobian matrix
, we developed a new regularization method in which the structure of data fidelity is incorporated .
we also developed a motion artifact removal filter , that can be applied to the data before image reconstructions , by using a sub - matrix of the jacobian matrix .
after explaining the developed methods , we will describe experimental results showing that the proposed fidelity - embedded regularization ( fer ) method combined with the motion artifact removal filter provides stable image reconstructions with satisfactory image quality even for very large regularization parameter values , thereby making the method irrelevant to the choice of the parameter value .
to explain the eit image reconstruction problem clearly and effectively , we restrict our description to the case of a 16-channel eit system for real - time time - difference imaging applications .
the sixteen electrodes ( @xmath0 ) are attached around a chosen imaging slice , denoted by @xmath1 .
we adopt the neighboring data collection scheme , where the device injects current between a neighboring electrode pair @xmath2 and simultaneously measures the induced voltages between all neighboring pairs of electrodes @xmath3 for @xmath4 . here
, we denote @xmath5 .
let @xmath6 be the electrical conductivity distribution of @xmath1 , and let @xmath7 denote the boundary surface of @xmath1 .
the electrical potential distribution corresponding to the @xmath8th injection current , denoted by @xmath9 , is governed by the following equations : @xmath10 where @xmath11 is the outward unit normal vector to @xmath12 , @xmath13 is the electrode contact impedance of the @xmath14th electrode @xmath15 , @xmath16 is the potential on @xmath15 subject to the @xmath8th injection current , and @xmath17 is the amplitude of the injection current between @xmath18 and @xmath19 . assuming that @xmath20 , the measured voltage between the electrode pair @xmath3 subject to the @xmath8th injection current at time @xmath21 is expressed as : @xmath22 the voltage data @xmath23 in a clinical environment are seriously affected by the following unwanted factors@xcite : * unknown and varying contact impedances making the data @xmath24 for the @xmath8th injection current unreliable and * inaccuracies in the moving boundary geometry and electrode positions . here , we set @xmath25 and @xmath26 . among those sixteen voltage data for each injection
current , thirteen are measured between electrode pairs where no current is injected , that is , the normal components of the current density are zero . for these data ,
it is reasonable to assume @xmath27 to get @xmath28 in .
discarding the remaining three voltage data , which are sensitive to changes in the contact impedances , we obtain the following voltage data at each time : @xmath29^t\ ] ] the total number of measured voltage data is @xmath30 at each time .
the data vector @xmath31 reflects the conductivity distribution @xmath6 , body geometry @xmath1 , electrode positions @xmath32 , and data collection protocol .
inevitable discrepancies exist between the forward model output @xmath63 and the measured data @xmath51 due to modeling errors and measurement noise : the real background conductivity is not homogeneous , the boundary shape and electrode positions change with body movements and there always exist electronic noise and interferences .
motion artifacts are inevitable in practice to produce errors in measured voltage data and deteriorate the quality of reconstructed images@xcite . to investigate how the motion artifacts influence the measured voltage data , we take time - derivative to both sides of assuming the domain @xmath1 varies with time .
it follows from the reynolds transport theorem that @xmath64 where @xmath65 is the outward - normal directional velocity of @xmath7 .
note that the last term of is the voltage change due to the boundary movement . using the chain rule ,
the first term of the right - hand side of is expressed as @xmath66 it follows from the integration by parts and that @xmath67 here , we neglected the contact impedances underneath the voltage - sensing electrodes and approximated @xmath68 . from - , can be expressed as : @xmath69 note that becomes when the boundary does not vary with time ( @xmath70 ) .
we linearize by replacing @xmath44 with the computed potential @xmath71 in : @xmath72 after discretization , can be expressed as : @xmath73 where @xmath74 is given by @xmath75^t\ ] ] with @xmath76 .
compared to , has the additional term @xmath74 that is the ( linearized ) error caused by the boundary movement .
this term multiplied by the strong sensitivities on the boundary @xmath77 becomes a serious troublemaker , and can not be neglected in because the vectors @xmath78^t$ ] for @xmath79 have large magnitudes . to filter out the uncertain data @xmath74
related with motion artifacts from @xmath80 , we introduce the boundary sensitive jacobian matrix @xmath81 , which is a sub - matrix of @xmath59 consisting of all columns corresponding to the triangular elements located on the boundary .
the boundary movement errors in the measured data @xmath80 are , then , assumed to be in the column space of @xmath81 .
the boundary errors are extracted by @xmath82 where @xmath83 is a regularization parameter and @xmath84 is the identity matrix .
then , the motion artifact is filtered out from data by subtraction @xmath85 .
the proposed motion artifact removal is performed before image reconstruction using any image reconstruction method . in this paper , the filtered data @xmath86 were used in places of @xmath80 for all image reconstructions .
severe instability arises in practice from the ill - conditioned structure of @xmath59 when some form of its inversion is tried . to deal with this fundamental difficulty , the regularized least - squares data - fitting approach
is commonly adopted to compute @xmath87 with a suitably chosen regularization parameter @xmath88 and regularization operator @xmath89 .
such image reconstructions rely on the choice of @xmath88 ( often empirically determined ) and @xmath89 using _ a priori _ information , suffering from over- or under - regularization .
we propose the fidelity - embedded regularization ( fer ) method : @xmath90 where the regularization operator @xmath91 is the diagonal matrix such that @xmath92 to explain the fer method , we closely examine the correlations among column vectors of the sensitivity matrix @xmath59 , described in fig .
[ figs : sensitivitymatrix ] .
the correlation between @xmath62 and @xmath93 can be expressed as @xmath94 where @xmath95 denotes the standard inner product . here
, @xmath96 is a solution of @xmath97 where @xmath98 is the characteristic function having @xmath99 on @xmath55 and @xmath100 otherwise .
the identity follows from @xmath101 for @xmath102@xcite .
this shows that the column vector @xmath62 is like an eeg ( electroencephalography ) data induced by dipole sources with directions @xmath103 at locations @xmath55 .
given that two dipole sources at distant locations produce mutually independent data , the correlation between @xmath104 and @xmath105 decreases with the distance between @xmath55 and @xmath106 .
[ figs : sensitivitymatrix ] shows a few images of the correlation @xmath107 as a function of @xmath108 for four different positions @xmath55 .
the correlation decreases rapidly as the distance increases . in the green regions where the correlation is almost zero
, @xmath93 is nearly orthogonal to @xmath109 .
[ figs : sensitivitymatrix ] shows that if @xmath55 and @xmath106 are far from each other , the corresponding columns of the sensitivity matrix are nearly orthogonal .
this somewhat orthogonal structure of the sensitivity matrix motivates an algebraic formula that directly computes the local ensemble average of conductivity changes at each point using the inner product between changes in the data and a scaled sensitivity vector at that point : @xmath110 where @xmath111 is the weighted average conductivity at the @xmath56th element @xmath55 and the weight is expressed in terms of the correlations between columns of @xmath59 .
it turns out that this simple formula shows a remarkable performance in terms of robustness , but requires a slight compromise in spatial resolution .
substituting @xmath112 into , the relation between @xmath113 and @xmath114 can be expressed as the following convolution form : @xmath115 where @xmath116 .
the non - zero scaling factor @xmath117 is designed for normalization .
the kernel @xmath118 satisfies the following : * @xmath119 for each @xmath56 , due to the non - zero scaling factor .
* @xmath118 decreases as the distance between @xmath55 and @xmath106 increases ( except near boundary where strong sensitivity arises ) .
hence , @xmath118 roughly behaves like a blurred version of the dirac delta function .
this is the reason why the formula directly computes the local ensemble average of conductivity changes at each point .
the algebraic formula can be seen as a regularized least - squares data - fitting method when the regularization operator is @xmath91 .
then , the formula can be expressed using @xmath91 in as @xmath120 this can be formulated similarly as for an extremely large value of @xmath88 ( @xmath121 ) when @xmath122 : @xmath123 here , @xmath124 is a devised scaling term to prevent the reconstructed image from becoming zero when @xmath88 goes to infinity .
the fer method in was proposed based on and .
when the regularization parameter @xmath88 is small ( @xmath125 ) , the fer method is equivalent to the regularized least - squares data - fitting method .
when @xmath88 is large ( @xmath126 ) , it converges to the algebraic formula and directly recovers the weighted average conductivity @xmath113 .
the regularization operation @xmath91 fully exploits the somewhat orthogonal structure of the sensitivity matrix , thereby embedding data fidelity in the regularization process .
adopting this carefully designed @xmath91 , the fer method provides stable conductivity image reconstructions with high fidelity even for very large regularization parameter values .
we applied the fer method to experimental data to show its performance .
we acquired the boundary geometry and electrode positions as accurate as possible to reduce forward modeling uncertainties@xcite .
a handheld 3d scanner was used to capture the boundary shape of the thorax and electrode positions ( fig .
[ figs : eit_system ] ) .
then , we set the electrode plane as the horizontal cross - section of the 3d - scanned thorax containing the attached electrodes ( fig .
[ figs : eit_system ] ) . the finite element method was employed to compute the sensitivity matrix @xmath59 by discretizing the imaging slice . here
, we used a mesh with 12,001 nodes and 23,320 triangular elements for subject a and a different mesh with 13,146 nodes and 25,610 triangular elements for subject b. figs .
[ fig : subjecta ] and [ fig : subjectb ] compare the performance of the proposed fer method in with the standard regularized least - squares method ( when @xmath127 is the identity matrix ) .
the regularization parameter of the standard method was heuristically chosen for its best performance , and the parameter of the fer method was set to be one of three different values @xmath128 .
the injection current was 1 ma@xmath129 at 100 khz , and the frame rate was 9 frames per second .
the reference frame at @xmath130 was obtained from the maximum expiration state .
the measured data , @xmath131 , represent the voltage differences between each time @xmath132 and @xmath130 .
the blue regions , which denote where conductivity decreased by inhaled air , increased during inspiration and decreased during expiration .
the fer method with @xmath121 was clearly more robust than the standard method that produced more artifacts originated from the inversion process . ' '' '' & @xmath130 & @xmath133 & @xmath134&@xmath135&@xmath136&@xmath137&@xmath138&@xmath139&@xmath140&@xmath141 & + & & & & & & & & & & & + & @xmath142 & @xmath143 & @xmath144&@xmath145&@xmath146&@xmath147&@xmath148&@xmath149&@xmath150&@xmath151 & + & & & & & & & & & & & + ' '' '' & @xmath130 & @xmath133 & @xmath134&@xmath135&@xmath136&@xmath137&@xmath138&@xmath139&@xmath140&@xmath141 & + & & & & & & & & & & & + & @xmath142 & @xmath143 & @xmath144&@xmath145&@xmath146&@xmath147&@xmath148&@xmath149&@xmath150&@xmath151 & + & & & & & & & & & & & + ' '' '' & @xmath130 & @xmath133 & @xmath134&@xmath135&@xmath136&@xmath137&@xmath138&@xmath139&@xmath140&@xmath141 & + & & & & & & & & & & & + & @xmath142 & @xmath143 & @xmath144&@xmath145&@xmath146&@xmath147&@xmath148&@xmath149&@xmath150&@xmath151 & + & & & & & & & & & & & + ' '' '' & @xmath130 & @xmath133 & @xmath134&@xmath135&@xmath136&@xmath137&@xmath138&@xmath139&@xmath140&@xmath141 & + & & & & & & & & & & & + & @xmath142 & @xmath143 & @xmath144&@xmath145&@xmath146&@xmath147&@xmath148&@xmath149&@xmath150&@xmath151 & + & & & & & & & & & & & + ' '' '' & @xmath130 & @xmath152 & @xmath137&@xmath153&@xmath142&@xmath154&@xmath147&@xmath155&@xmath156&@xmath157 & + & & & & & & & & & & & + & @xmath158 & @xmath159 & @xmath160&@xmath161&@xmath162&@xmath163&@xmath164&@xmath165&@xmath166&@xmath167 & + & & & & & & & & & & & + ' '' '' & @xmath130 & @xmath152 & @xmath137&@xmath153&@xmath142&@xmath154&@xmath147&@xmath155&@xmath156&@xmath157 & + & & & & & & & & & & & + & @xmath158 & @xmath159 & @xmath160&@xmath161&@xmath162&@xmath163&@xmath164&@xmath165&@xmath166&@xmath167 & + & & & & & & & & & & & + ' '' '' & @xmath130 & @xmath133 & @xmath134&@xmath135&@xmath136&@xmath137&@xmath138&@xmath139&@xmath140&@xmath141 & + & & & & & & & & & & & + & @xmath158 & @xmath159 & @xmath160&@xmath161&@xmath162&@xmath163&@xmath164&@xmath165&@xmath166&@xmath167 & + & & & & & & & & & & & + ' '' '' & @xmath130 & @xmath133 & @xmath134&@xmath135&@xmath136&@xmath137&@xmath138&@xmath139&@xmath140&@xmath141 & + & & & & & & & & & & & + & @xmath158 & @xmath159 & @xmath160&@xmath161&@xmath162&@xmath163&@xmath164&@xmath165&@xmath166&@xmath167 & + & & & & & & & & & & & + note that the degree of orthogonality of the columns of the sensitivity matrix depends on the current injection pattern .
this makes the performance of the fer method depend on the current injection pattern since it incorporates the structure of the sensitivity matrix in the regularization process .
for example , if we inject currents between diagonal pairs of electrodes , the corresponding sensitivity matrix becomes less orthogonal compared with that using the neighboring current injection protocol , thus producing more blurred images . in contrast , a more narrower injection angle , for example , using a 32-channel eit system with the adjacent injection pattern , can enhance the orthogonality of the corresponding sensitivity matrix .
however , the narrower injection angle results in poor distinguishability@xcite and may deteriorate the image quality . to maximize the performance of the fer method , balancing between orthogonality of the sensitivity matrix and distinguishability
should be considered when designing a data collection protocol .
the direct algebraic formula or for @xmath121 can be expressed as a transpose of a scaled sensitivity matrix .
this type of direct approach was suggested in the late 1980s by kotre@xcite , but was soon abandoned owing to poor performance and lack of theoretical grounding@xcite . since then
, regularized inversion of the sensitivity matrix has been the main approach for eit image reconstruction .
kotre s method using the normalized transpose of @xmath168 was regarded as an extreme version of the backprojection algorithm in eit@xcite , in the sense of the radon transform in ct ; in the case when @xmath168 is the radon transform of ct , its adjoint is known as the backprojector . with this interpretation
, it seems that kotre s method is very sensitive to forward modeling uncertainties . in the fer method
, @xmath118 is a scaled version of the adjoint of the jacobian , which can be viewed as a blurred version of the dirac delta function .
the fer method with @xmath121 becomes a direct method for robust conductivity image reconstructions without inverting the sensitivity matrix @xmath59 .
since the 1980s , many eit image reconstruction methods have been developed to overcome difficulties in achieving robust and consistent images from patients in clinical environments .
recent clinical trials of applying eit to mechanically ventilated patients have shown its feasibility as a new real - time bedside imaging modality
. they also request , however , more robust image reconstructions from patients data contaminated by noise and artifacts .
the proposed fer method achieves both robustness and fidelity by incorporating the structure of the sensitivity matrix in the regularization process . unlike most other algorithms ,
the fer method also offers direct image reconstructions without matrix inversion .
this has a practical advantage especially in clinical environments since the direct method does not require any adjustment of regularization parameters .
in addition to time - difference conductivity imaging , the fer method enables robust spectroscopic admittivity imaging of both conductivity and permittivity , and can employ frequency - difference approaches as well . although we showed that @xmath113 in provides a satisfactory approximation to @xmath114 , it is difficult to estimate its accuracy
. this is related to the kernel @xmath118 and the structure of the sensitivity matrix @xmath59 in which the current injection pattern is incorporated .
it is a challenging issue to mathematically characterize how precisely the kernel @xmath118 approximates the dirac delta function with respect to the current inject pattern .
k.l . and j.k.s .
were supported by the national research foundation of korea ( nrf ) grant 2015r1a5a1009350 .
e.j.w . was supported by a grant of the korean health technology r&d project ( hi14c0743 ) .
t. meier et al . , assessment of regional lung recruitment and derecruitment during a peep trial based on electrical impedance tomography " , _ intensive care med . _ ,
543 - 550 , 2008 .
i. frerichs et al . , chest electrical impedance tomography examination , data analysis , terminology , clinical use and recommendations : concesus statement of the translational eit development study group " , _ thorax _ at press , 2016 .
d. isaacson , distinguishability of conductivities by electric current computed tomography " , _ ieee trans . med .
imaging _ ,
mi-5 , nol .
2 , pp . 91 - 95 , 1986 .
b. h. brown , d. c. barber , and a. d. seagar , applied potential tomography : possible clinical applications " , _ clin .
109 - 121 , 1985 f. santosa and m. vogelius , a back - projection algorithm for electrical impedance imaging " , _
siam j. appl .
216 - 243 , 1990 m. cheney , d. isaacson , j. newell , s. simske , and j. goble , noser : an algorithm for solving the inverse conductivity problem " , _ internat .
j. imaging systems and technology _ , vol .
66 - 75 , 1990 .
s. siltanen , j. mueller , and d. isaacson , an implementation of the reconstruction algorithm of a nachman for the 2d inverse conductivity problem " , _ inverse problems _ , vol .
681 - 699 , 2000 .
j. l. mueller , s. siltanen , d. isaacson , a direct reconstruction algorithm for electrical impedance tomography " , _ ieee trans . med .
21 , no . 6 , pp .
555 - 559 , 2002 .
m. k. choi , b. bastian , and j. k. seo , regularizing a linearized eit reconstruction method using a sensitivity - based factorization method " , _ inverse probl .
22 , no . 7 , pp . 1029 - 1044 , 2014 .
v. kolehmainen , m. vauhkonen , p. a. karjalainen , j. p. kaipio , assessment of errors in static electrical impedance tomography with adjacent and trigonometric current patterns " , _ physiol .
4 , pp . 289 - 303 , 1997 . | electrical impedance tomography ( eit ) provides functional images of an electrical conductivity distribution inside the human body . since the 1980s
, many potential clinical applications have arisen using inexpensive portable eit devices .
eit acquires multiple trans - impedance measurements across the body from an array of surface electrodes around a chosen imaging slice .
the conductivity image reconstruction from the measured data is a fundamentally ill - posed inverse problem notoriously vulnerable to measurement noise and artifacts .
most available methods invert the ill - conditioned sensitivity or jacobian matrix using a regularized least - squares data - fitting technique .
their performances rely on the regularization parameter , which controls the trade - off between fidelity and robustness . for clinical applications of eit ,
it would be desirable to develop a method achieving consistent performance over various uncertain data , regardless of the choice of the regularization parameter . based on the analysis of the structure of the jacobian matrix , we propose a fidelity - embedded regularization ( fer ) method and a motion artifact removal filter . incorporating the jacobian matrix in the regularization process , the new fer method with the motion artifact removal filter offers stable reconstructions of high - fidelity images from noisy data by taking a very large regularization parameter value .
the proposed method showed practical merits in experimental studies of chest eit imaging . |
high - energy electro - disintegration of the deuteron is a powerful tool to investigate , first , how and when at larger energies and momentum transfers the description in terms of hadrons ( nucleons and mesons ) transforms into the one in terms of quarks and gluons and , second , the colour transparency ( ct ) effects predicted by qcd . in terms of relevant feynman diagrams , one expects that whereas at comparatively low energies virtual nucleons and mesons with standard propagators can be used , at large enough virtualities such a description gradually becomes invalid . to clearly see this phenomenon one has to be able to reach high enough virtualities in the process , which can be achieved by choosing the kinematics forbidden for free ( on - mass - shell ) nucleons ,
i.e. the so - called _ cumulative _ kinematics , which corresponds to values of the bjorken scaling variable larger than one ( note that the importance of studying the @xmath5-dependence of ct was first stressed in ref .
@xcite ) . in the pwia
the cross - sections are directly related to the behaviour of the deuteron wave function .
however such a direct relation is broken by fsi , which naively are expected to be large in the cumulative region ; it is here that ct effects are to be taken into account .
ct predicts that at high @xmath8 fsi become small , so that , in principle , choosing both @xmath5 and @xmath8 large enough one can neglect fsi and have access , via the pwia , to the high - momentum behaviour of the nuclear wave function .
it is of particular importance that for high values of @xmath5 and @xmath8 the rescattering energy may be quite small @xcite , so that only two - nucleon intermediate states can be considered and the simplest feynman diagrams with only nucleon lines ( although at high virtuality ) have to be evaluated . in such a kinematical situation
ct effects can originate from the virtual excitation of the ejectile . in this study
we investigate the relative role of fsi and ct effects in the deuteron electro - disintegration , using an approximate picture in which the spins are altogether averaged out . taking the relativistic spin into account , apart from purely technical difficulties
, inevitably introduces a variety of unknown off - shell components of both the electromagnetic production vertex and the rescattering amplitude , which deprives the results of any predictive power . to relate our absolute cross - sections to the experimental data we introduce an effective electromagnetic form factor , chosen to reproduce the pwia results with full relativistic spins @xcite . in our approach ct
is introduced on the hadronic level using its equivalence to the final formation time ( fft ) effect @xcite .
separating the leptonic part , the process we are going to study is ^*(q)+d(2p)n(p_1)+n(p_2 ) , [ process1 ] where 4-momenta are denoted in brackets . the total c.m .
energy squared in the process is @xmath9 where @xmath10 is the deuteron mass .
the cumulative region we are going to study corresponds to @xmath11 from eq .
( [ esse ] ) one concludes that at @xmath5 close to 2 the c.m .
energy remains close to the threshold even if @xmath8 is large , so that only the lowest two - nucleon intermediate states can be considered .
the calculation of the cross - section for the process ( [ process1 ] ) then reduces to the evaluation of the square modulus of the two standard diagrams shown in fig .
1 , which correspond to the pwia and rescattering contributions , respectively . 3.5 in our final expression for the inclusive process e(l)+d(2p)e(l)+x [ process2 ] can be standardly expressed through the hadronic tensor @xmath12 , which is just the square of the two amplitudes shown in fig.1 integrated over the momenta of the two final nucleons .
it can be shown that in order to obtain the exclusive cross - section for the lepto - disintegration , averaged over the azimuthal angles , it is sufficient to drop this integration .
the pwia contribution to the amplitude is just the hadronic matrix element of the electromagnetic current j_^pwia=2(q^2)k_g(k_1 ^ 2 ) , [ incl ] where @xmath13 is the effective electromagnetic form - factor of the ( scalar ) nucleon , the vector @xmath14 is chosen to guarantee conservation of the electromagnetic current , and @xmath15 is the momentum of the active nucleon before the interaction .
the quantity @xmath16 is the relativistic deuteron wave function , which in principle can be sought as a solution of the bethe - salpeter equation for the deuteron .
however the relativistic potential for this equation is unknown at high momentum transfers and virtualities , so rather than use forms for this potential fitted to comparatively low energy and momentum transfer scattering , we directly approximate @xmath17 using as a guide its non - relativistic limit . to calculate the rescattering contribution we choose a system with @xmath18 ( `` lab '' system for reaction ( [ process1 ] ) ) which simplifies the integrations over the intermediate nucleon momenta . in this system
the integration over @xmath19 puts either the spectator or the active nucleon on the mass - shell , depending on whether @xmath20 is lower or higher than @xmath21 . in both cases , as in the pwia contribution , the integrand involves only the relativistic deuteron wave function with one of the nucleons on its mass shell . as a result
we find the rescattering contribution to the matrix element of the hadronic current in the form j_^resc=2(q^2 ) dv(k_1,k_2)k_g(v ) , [ jresc ] where @xmath22 , @xmath23 and @xmath24 is the rescattering amplitude , with @xmath25 and @xmath26 having different forms in the two mentioned regions of integrations .
all invariant arguments entering the integrand in eq .
( [ jresc ] ) have to be expressed through the light - cone integration variables , taking into account that either @xmath27 , or @xmath28 .
the total hadronic tensor is obtained as w_=dv(p_1,p_2)(j_^pwia+j_^resc ) ( j_^pwia+j_^resc)^ * , [ wtot ] where @xmath29 is the standard invariant phase volume for the produced nucleons .
as far as ct is concerned , we introduce it via the fft of the hit hadron , which manifests itself through a dependence of the scattering amplitudes and verteces on the virtuality of the hit hadron after @xmath30 absorption ( see [ 3 ] ) .
we assume that the effect of this dependence can be modelled by a monopole form - factor , which is equivalent to changing the ejectile propagator as follows - .
[ denom ] this substitution is equivalent to assuming that there are two different ejectile states with masses @xmath31 and @xmath32 whose contribution to rescattering cancels out in the limit of high momenta , in agreement with the underlying ideas of ct [ 3 ] . following refs .
@xcite , @xcite , we choose for the mass @xmath32 the value 1.8 gev .
3.0 in we parametrize the relativistic deuteron wave function @xmath33 using as a guide the nucleon density in the deuteron at comparatively low momenta . in the non - relativistic limit one obtains g^2(k^2)=2m(2)^3|(*k*^2)| ,
* k*^2=(m^2-k^2-m ) , where @xmath34 is the deuteron binding energy . for @xmath35 we have taken the form which corresponds to the av14 interaction @xcite .
as for the rescattering amplitude , it was chosen in the form a(s , t)=(+i)^tot(s)e^bt , with the values of the parameters @xmath36 , @xmath37 and @xmath38 taken from @xcite .
as already mentioned , our main goal was to estimate the magnitude of the fsi and the influence of ct ( or , equivalently , fft ) in the cumulative region .
accordingly , our basic quantities to be calculated are the ratios of the rescattering to the pwia contributions . we expect that the error due to neglecting spins is reduced in these ratios
. still , to have a clearer physical picture and to be able to compare with the experimental data , we also tried to calculate absolute values of the cross - sections . to this end
we introduced an effective electromagnetic form - factor for the active nucleon , chosen to approximately take into account magnetic interaction of both the proton and neutron .
comparison of our impulse approximation with the results obtained with a full account of spins [ 2 ] leads to the choice ^2(q^2)=_d^2(q^2 ) , [ formfactor ] where @xmath39 is the standard dipole form - factor , @xmath40 , and @xmath41 are the anomalous magnetic moments of the proton and neutron . with this choice
our pwia results practically coincide with the ones with full relativistic spins taken into account .
the inclusive cross - sections for the process ( [ process2 ] ) have been calculated at points corresponding to the experimental data of @xcite , with the initial electron energy @xmath42 gev and scattering angle @xmath43 .
we have considered values for the final electron energy which cover the region of @xmath5 in the interval @xmath44 .
some relevant kinematical values characterizing the chosen points are listed in table 1 .
3.5 in -10 mm -1.5 in .some kinematical variables in the inclusive electro - disintegration cross section corresponding to the kinematics of of @xcite , _ viz _ incident electron energy @xmath45 and scattering angle @xmath46 .
@xmath47 is the energy transfer , @xmath5 the bjorken scaling variable , @xmath8 the square 4-momentum transfer , @xmath48 the produced invariant mass , and @xmath49 the the momentum of the struck nucleon in the system where the spectator is at rest @xmath49 .
note that the inelastic threshold corresponds to @xmath50 ( @xmath51 ) ( cf .
[ cols="^,^,^,^,^ " , ] [ table1 ] it can be seen , that at high values of @xmath5 ( high cumulativity ) @xmath49 is small , well below the threshold energy for pion production .
this justifies our approach based on the assumption that only nucleon degrees of freedom are relevant ; however at smaller @xmath5 @xmath49 grows and such an assumption becomes of disputable validity . 2.5 in in fig .
2 the ratios of the full cross section ( which includes fsi with or without fft ) to the pwia cross section is shown .
it can be seen that the rescattering contribution , which is very small at @xmath52 , steadily grows with @xmath5 , reaching an order of about 50% already at @xmath53 .
the relative role of fft ( or ct ) also rises with @xmath5 . at
the maximum value of @xmath5 studied , @xmath54 , fft effects decrease fsi by @xmath7% .
this however does not make fsi smaller than the pwia contribution , so that they can not be neglected at all .
let us now discuss the exclusive cross - section @xmath55 . in order to minimize the error of neglecting spins
, we consider the reduced cross - section n_eff(|*p*_2| ) n_eff(|*p*_2|,x , q^2)= momentum distribution in the deuteron . here
@xmath56 is the momentum of the unobserved nucleon ( the missing momentum ) .
3 shows the effective momentum distributions _ vs _ the missing momentum at @xmath57 , calculated taking rescattering and fft into account .
calculations have been performed at fixed values of @xmath5 and @xmath8 and different missing momenta @xmath58 . due to energy conservation
, this corresponds to different values of the angle between @xmath59 and @xmath60 .
2.7 in -1 mm in figs .
4 and 5 the effects of rescattering and fft are shown separately at @xmath57 , and in the deep kinematical region , @xmath61 .
it can be seen from fig.4 that at @xmath57 fft effects , as expected , increase with @xmath8 , in agreement with the results obtained for the process @xmath62 ( @xcite ) . at @xmath61 ( fig .
5 ) one observes that with the growth of the missing momentum , fsi appreciably decrease due to fft , which makes the distorted cross - section more similar to the pwia one .
using the feynman diagram technique we have calculated the rescattering contribution to the cross - section of inclusive and exclusive deuteron electro - disintegration , paying particular attention to the cumulative region @xmath2 .
our main goal has been to study ct effects , which we have introduced via the fft of the struck nucleon . in our calculations spins
has been averaged out and all particles were treated as scalar ones . the error introduced by this approximation
is expected to cancel to a large extent in the ratio of the total cross - section ( which includes fsi and fft effects ) , with and without fsi and fft effects , to the pwia cross section . to be able to compare absolute magnitudes of the cross - sections with the experimental data we used an effective electromagnetic vertex for the active nucleon , chosen to describe the pwia results with spins .
the results of our calculations show that the relative magnitude of the fsi steadily grows with @xmath5 . at @xmath63 ( gev / c)@xmath4 and @xmath64
, the rescattering contribution raises the cross - section nearly by a factor of 2.5 , whereas the introduction of fft decreases back the cross section by a factor of 1.6 , which is evidently not enough to disregard fsi altogether .
as expected , at large values of @xmath5 the influence of fft grows up : it can be seen from fig .
5 that at @xmath65 ( gev / c)@xmath4 and @xmath66 gev / c , fft decreases the pure fsi results by a factor of 4 . in conclusion , we have found that the ct or fft effects are clearly visible in the electro - disintegration of the deuteron in the cumulative region , which may serve as a tool for their experimental study .
we have also found that although fft effects decrease the effects of the fsi rather substantially at @xmath67 , this is by far not sufficient to neglect fsi altogether .
this work was partially supported by the ministero dellistruzione , universit e ricerca ( miur ) , through the funds cofin01 . m. b. and l.p.k .
are indebted to the university of perugia and infn , sezione di perugia , for warm hospitality and financial support . 100 b.k .
jennings and b.z .
kopeliovich , phys .
* 70 * ( 1993 ) 3384 c. ciofi degli atti , l.p .
kaptari and d. treleani , phys.rev . *
c63 * ( 2001 ) 044601 m. a. braun , c. ciofi degli atti and d. treleani , phys . rev .
* c62 * ( 2000 ) 034606 l.l .
frankfurt , w.r .
greenberg , g.a .
miller and m.i .
strikman , phys .
, * c46 * ( 1992 ) 2547 r.b .
wiringa , r. a. smith and t. l. ainsworth , phys .
rev . * c29 * ( 1984 ) 1207 r. a. arndt et al . , `` partial - wave analysis facility ( said ) '' , http:/said.phys.vt.edu s. rock _
et al _ , phys .
lett.*49 * ( 1982 ) 1139 ; + r. g. arnold _ et al _ , _ ibid . _ * 61*(1988 ) 806 h. morita , m. a. braun , c. ciofi degli atti , and d. treleani , nucl .
a699 * ( 2002 ) 328c | the role of rescattering due to the final state interaction ( fsi ) and the influence of the finite formation time ( fft ) on the inclusive @xmath0 and exclusive @xmath1 electro - disintegration of the deuteron are studied in the cumulative kinematical region @xmath2 and moderate values of the 4-momentum transfer @xmath3 ( gev / c)@xmath4 .
the spins are averaged out .
it is found that in the inclusive process the relative magnitude of rescattering steadily grows with @xmath5 and that at @xmath6 it has the same order as the plane wave impulse approximation ( pwia ) contribution , with the finite formation time effects decreasing the rescattering contribution by @xmath7% . in the exclusive process , with increasing momentum transfer
, fft substantially reduces the effects from fsi , although the latter are still appreciable in the region of momentum transfer investigated . |
in this paper we reexamine and analyze the phenomenon of `` bistability '' that occurs in chemical models of dark interstellar clouds .
bistability was first discussed by le bourlot et al .
( 1993 ) , who found that multiple solutions to the chemical rate - equations sometimes appear in numerical computations of the gas - phase abundances of atomic and molecular species for steady - state dark cloud conditions .
when bistability occurs , three sets , rather than a single unique set , of chemical abundances are predicted for identical input cloud parameters such as the total gas density , the cosmic - ray ( or x - ray ) ionization rate , and the gas - phase supply of heavy elements . under such circumstances , the solution that is converged to in a numerical simulation , and perhaps , the actual steady - state of a real physical system , depend on the initial conditions and history of the gas .
the subject of bistability begins with pineau des forets et al .
( 1992 ) who studied how the computed chemical states of dark clouds depend on the gas density and ionization rate .
they found , as would be expected , that for a given ionization rate the fractional ionization of the gas decreases roughly as the square - root of the gas density .
however , they also found that the ionization fraction and associated chemical composition changes abruptly , or quasi - discontinuously , at a specific critical transition density . below the critical density
the clouds are in what pineau des forets et al
. referred to as a `` high - ionization '' phase ( hip ) , characterized by high abundances of atomic ions . for densities greater than the critical density
the clouds are in a `` low - ionization '' phase ( lip ) , with high abundances of molecular ions .
pineau des forets et al . also identified a chemical instability that , they argued , is responsible for the abrupt transition that they encountered in their calculations .
bistability was subsequently discovered by le bourlot et al .
( 1993 ) who found that instead of a phase change at a specific critical density , the lip and hip `` branches '' can sometimes overlap for a narrow range of densities .
a third ( unstable ) branch also appears , intermediate between the lip and hip solutions .
further investigations have explored the effects of the assumed gas phase elemental abundances , ionization rates , gas temperatures , chemical reaction rate coefficients , and gas - grain interactions , on the existence and character of the multiple solutions ( le bourlot et al .
1995ab ; shalabiea & greenberg 1995 ; lee et al .
1998 ; pineau des forets & roueff 2000 ; viti et al . 2001 ; charnley & markwick 2003 ) .
these studies have shown that bistability is indeed related to the transition from the hip to lip ( or vice versa ) , and that it is sensitive to quantities such as the gas - phase heavy - element abundances ( le bourlot et al . 1995 ; viti et al .
2001 ) , and the adopted chemical network ( lee et al .
1998 ) , including specifically the dissociative recombination rate coefficient of h@xmath0 ( pineau des forets & roueff 2000 ) . in this paper
we explain why the hip and lip can overlap , and we identify the chemical mechanisms that allow the possibility of multiple solutions .
we will show that bistability can be simply understood and analyzed using just a handful of reactions . in a nutshell , when bistability occurs atomic ions ( usually s@xmath1 ions ) are the dominant positive charge carriers for both hip and lip conditions .
two modes of neutralization are then available to the gas . in the hip ,
neutralization proceeds by slow radiative recombination . in the lip ,
neutralization occurs by rapid dissociative recombination , via the formation of intermediate molecular ions in reactions with o@xmath2 molecules ( e.g. , @xmath3 ) .
bistability occurs when a low o@xmath2 abundance in the hip can be maintained up to maximal gas densities that are _ greater _ than the minimal densities above which a large o@xmath2 density is maintained in the lip .
we will show by means of a simple analysis why this is possible .
the explanation involves a chemical instability , of a type first introduced by pineau des forets et al .
( 1992 ) , but not given sufficient attention in subsequent studies .
bistability may be of primarily theoretical interest since , as we will discuss , it is a purely gas - phase chemical effect . if realistic gas - dust neutralization processes are included , the multiple solutions are damped or eliminated entirely ( cf . shalabiea & greenberg 1995 ) .
we will illustrate this by considering the effect of grain - assisted ion - electron recombination ( weingartner & draine 2001 ) on the chemistry .
nevertheless , because bistability has been the subject of considerable discussion in the literature , we feel that a fresh analysis of how , why , and when it occurs is warranted .
the plan of our paper is as follows . in
2 we present results of an illustrative numerical computation in which bistability occurs . in
3 we discuss the key chemical instability that drives the steady - state solutions to either the lip or hip . in
4 we derive simple analytic formulae that show how the range of gas densities and ionization rates for which the gas becomes bistable depends on the elemental abundances and several important chemical rate coefficients . in
5 we analyze the sensitivity of bistability to the h@xmath0 dissociative recombination rate coefficient . in 6 we explain why gas - grain neutralization processes damp and eliminate bistability .
we summarize in 7 .
we first illustrate the phenomenon of bistability in a detailed numerical computation . for this purpose
we calculate the steady - state atomic and molecular abundances for dark cloud conditions , as functions of the gas density .
we include only gas - phase formation and destruction processes , except for molecular hydrogen ( h@xmath2 ) , which we assume is formed on grain surfaces .
the chemistry is driven by cosmic - ray impact ionization of hydrogen and helium , followed by standard sequences of ion - molecule , charge - exchange , and neutral - neutral reactions .
the chemistry is also mediated by radiative and dissociative recombination , and cosmic - ray induced photoionization and photodissociation .
the effects of externally incident fuv photons are excluded .
we use the boger & sternberg ( 2005 ) network for a system consisting of 70 atomic and molecular carbon , nitrogen , oxygen , sulfur , and silicon bearing species ( see also sternberg & dalgarno 1995 ) .
we neglect all other heavy elements .
our set of 868 reactions and rate coefficients is based mainly on the umist99 compilation ( le teuff et al .
2000 ) with some modifications and alterations . as in boger & sternberg ( 2005 )
we adopt @xmath4 @xmath5 gas - phase abundances , with c / h=@xmath6 ( cardelli et al .
1993 ) , n / h=@xmath7 ( meyer et al .
1997 ) , o / h=@xmath8 ( meyer et al .
1998 ) , si / h=@xmath9 ( cardelli et al . 1994 ) , and s / h=@xmath10 ( lepp et al . 1988 ) .
we set he / h=0.1 . we assume a constant gas temperature of @xmath11 k for the entire range of gas densities that we consider . in our calculations
we solve the equations of chemical equilibrium @xmath12 for the densities @xmath13 of all species @xmath14 in our set . in these equations ,
@xmath15 are the temperature - dependent rate coefficients ( @xmath16 s@xmath17 ) for chemical reactions between species @xmath18 and @xmath19 that lead to the production of @xmath14 , and @xmath20 are the cosmic - ray destruction rates ( s@xmath17 ) of species @xmath18 with products @xmath14 .
the parameters @xmath20 include impact ionizations of hydrogen and helium by the cosmic - ray protons and secondary electrons , as well as photo - destruction of heavy atomic and molecular species by induced uv photons .
the @xmath20 are all proportional to @xmath21 s@xmath17 , the cosmic - ray ionization rate of molecular hydrogen ( sternberg et al . 1987 ; gredel et al .
1989 ) the densities @xmath13 must also satisfy the supplementary equations of element and charge conservation .
these are , @xmath22 where @xmath23 is the number of atoms of element @xmath24 contained in species @xmath14 , @xmath25 is the total abundance of element @xmath24 relative to the total density @xmath26 of hydrogen nuclei , and @xmath27 where @xmath28 is the charge of species @xmath14 . in our computations ,
the negative charge is carried entirely by free electrons .
h@xmath2 molecules are formed primarily on dust grains and are destroyed by cosmic - ray ionization . a separate equation
may then be written for the h / h@xmath2 balance , @xmath29 where @xmath30 and @xmath31 are the atomic and molecular hydrogen densities respectively , and @xmath32 @xmath33 s@xmath17 is the h@xmath2 grain surface formation rate coefficient ( biham et al . 2002 ; vidali et al .
2004 ) . for fixed gas temperature and gas - phase elemental abundances , the density fractions @xmath34
as determined by equations ( [ ceqns])-([heqn ] ) depend on the single parameter @xmath35 ( e.g. , lepp & dalgarno 1996 ; lee et al .
1998 ) , as is readily seen by dividing each term in these equations by @xmath36 .
we consider a range of parameters for which almost all of the hydrogen is in the form of h@xmath2 , so that @xmath37 .
we therefore present our results as functions of the ratio @xmath38 ( @xmath39 ) .
we solve equations ( [ ceqns])-([heqn ] ) numerically via newton - raphson ( nr ) iteration . in our procedure
we vary the density along the hip and lip branches and use the solutions from the previous density step as the `` initial guess '' for the nr iteration in the next density step . in the bistability region
we carry out repeated nr iterations with random initial guesses , so that all three branches are found . in figure 1
we plot the densities @xmath40 , and density fractions @xmath41 , as functions of @xmath38 ranging from 10@xmath42 to 10@xmath43 @xmath39 , for several species of interest .
panels ( a ) and ( b ) display our results for electrons , c@xmath1 , s@xmath1 , h@xmath0 , and a sum over all molecular ions . panels ( c ) and
( d ) display our results for the neutral species c , o , co and o@xmath2 .
several important features are apparent in figure 1 , some of which have been noted in previous discussions . for our assumed elemental abundances and gas temperature , a high - ionization phase ( hip ) is present for densities less
than @xmath44 @xmath39 , and a low - ionization phase ( lip ) is present for densities greater than @xmath45 @xmath39 .
the hip and lip overlap , and the gas is bistable , between these minimal and maximal densities .
the hip and lip solutions are connected by a third branch that we refer to here as the mix solution . in the hip ,
the positive charge is carried entirely by atomic ions .
s@xmath1 is the dominant ion down to @xmath46 @xmath39 . at lower densities c@xmath1
takes over , and for @xmath47 @xmath39 , h@xmath1 dominates . in the lip , the molecular ions hco@xmath1 and h@xmath48o@xmath1 are the primary positive charge carriers for densities greater than @xmath49 @xmath39
however , s@xmath1 dominates at lower densities in the lip .
most importantly , in our calculation s@xmath1 is the primary positive charge carrier in _ both _ the hip and lip for the densities at which bistability occurs . the fractional ionization increases from a minimum value @xmath50 as the density decreases from @xmath51 in the hip . in the lip ,
the fractional ionization decreases from a maximum value of @xmath52 as the density increases from @xmath53 .
the fractional ionization drops by a factor of @xmath54 in moving from the hip to lip in the bistability region . the free atomic carbon density and the c / co abundance ratio is large in the hip . the ratio @xmath55 in the bistability region , and exceeds unity for @xmath56 @xmath39 . in the lip the c density and the c / co ratio are small . in our computation the c / co ratio drops by a factor @xmath57 in moving from the hip to the lip .
the drop in c / co is associated with an _ increase _ in the o@xmath2 density and the o@xmath2/co ratio .
it is very significant that the c and o@xmath2 densities vary in opposite directions in the transition between the hip and lip . on the mix branch
the fractional ionization , and the c / co and o@xmath2/co ratios , are intermediate between the hip and lip .
s@xmath1 is also the dominant positive ion on the mix branch .
however , the fractional ionization increases with gas density along the mix . in table 1 we list the densities of selected atomic and molecular species for the three hip , lip , and mix solutions , for @xmath58 @xmath39 and @xmath59 s@xmath17 .
we also list the hip / lip density ratios for these values of @xmath60 and @xmath4 .
atomic ions , e.g. , s@xmath1 , si@xmath1 , and c@xmath1 , are abundant in the hip solution . in the lip solution , molecular ions such as h@xmath0 , hco@xmath1 , and h@xmath48o@xmath1 are abundant , leading to more oh , h@xmath2o , and o@xmath2 than in the hip . the oxygen
bearing molecules , no , so , and sio , are enhanced by the large o@xmath2 density that is maintained in the lip .
carbon bearing molecules , such as cn and hcn , are enhanced in the hip because of the large atomic carbon density in this phase ( see also boger & sternberg 2005 ) .
cs is produced by reactions of so ( enhanced in the lip ) with c ( enhanced in the hip ) so that the cs density changes by only a factor of @xmath61 in moving from the lip and hip .
figure 1 shows that as the gas density @xmath60 is increased along the sequence of hip solutions an abrupt jump to the lip branch occurs when @xmath60 rises above @xmath51 .
similarly , as @xmath60 is decreased along the lip , a jump to the hip occurs when @xmath60 falls below @xmath53 .
as we now discuss , these jumps are signatures of a simple chemical instability involving electrons , s@xmath1 and h@xmath0 ions , c atoms , and o@xmath2 molecules .
the instability can be understood by considering the gas - phase formation and destruction sequences for these species , for gas densities where the bistable solutions occur .
we first discuss the hip and lip , and we then consider the mix .
we begin with h@xmath0 . in both the hip and lip ,
this ion is formed by cosmic - ray ionization of h@xmath2 followed by proton transfer , @xmath62 @xmath63 in the hip , the electron fraction is large , and dissociative recombination @xmath64 @xmath65 dominates the removal of the h@xmath0 . however , in the lip the electron fraction is small , and the h@xmath0 is removed by @xmath66 for a gas temperature of 50 k , dissociative recombination dominates the destruction of h@xmath0 when @xmath67 . in both
the hip and lip , the production of o@xmath2 is initiated by proton transfer @xmath68 which is followed by abstraction @xmath69 @xmath70 dissociative recombination @xmath71 @xmath72 and the neutral - neutral reaction @xmath73 however , the o@xmath2 is destroyed in different ways in the two phases . in the hip ,
the density of free atomic carbon is large , and the o@xmath2 is removed by @xmath74 in the lip , the c density is small , and the o@xmath2 is removed much less efficiently in reactions with less abundant atoms and atomic ions .
an upper limit on the o@xmath2 density in the lip is set by cosmic - ray induced photodissociation @xmath75 reaction ( r13 ) is inefficient , and the o@xmath2 density becomes large in the lip . in the hip ,
c@xmath1 is produced by cosmic - ray driven helium impact ionization of co , @xmath76 @xmath77 this is then followed by rapid charge transfer @xmath78 which leads to the production of s@xmath1 and c. the s@xmath1 ions are removed by radiative recombination , @xmath79 the c atoms are removed by cosmic - ray induced photoionization @xmath80 because cosmic - ray photoionization is inefficient , a large c density is maintained in the hip . in the lip ,
c@xmath1 is also produced by helium impact ionization ( [ r14 ] and [ r15 ] ) . however , because the o@xmath2 density is large in the lip , the c@xmath1 ions are rapidly neutralized by the sequence @xmath81 @xmath82 and the c@xmath1 density becomes small . in the lip , charge transfer between c@xmath1 and s ( [ r16 ] ) contributes negligibly to the removal of c@xmath1 , or to the formation of s@xmath1 .
the main source of s@xmath1 in the lip is cosmic - ray induced photoionization @xmath83 crucially , o@xmath2 also dominates the neutralization of s@xmath1 in the lip , via @xmath84 @xmath85 for a gas temperature of 50 k , neutralization of s@xmath1 by o@xmath2 dominates over radiative recombination when @xmath86 .
finally , the c atoms formed by ( r20 ) are also removed by reactions with o@xmath2 ( [ r12 ] ) , and the c density becomes small in the lip . for the gas densities of interest ,
most of the gas - phase sulfur is present as s and s@xmath1 in both the hip and lip , with only a small fraction bound in molecules .
s@xmath1 is the primary positive charge carrier in both the hip and lip .
however , in the hip s@xmath1 is removed by radiative recombination , whereas in the lip by reactions with o@xmath2 . in the hip ,
h@xmath0 is removed by dissociative recombination , c is removed by cosmic - ray induced photoionization , and o@xmath2 is removed by c. in the lip , h@xmath0 is removed by co , c is removed by o@xmath2 , and o@xmath2 is limited by cosmic - ray induced photodissociation .
the abundance ratio c / o@xmath2 is large in the hip , and is small in the lip .
now consider a gas where s@xmath1 is again the dominant charge carrier , but which differs from the hip or lip in that dissociative recombination contributes significantly to the removal of h@xmath0 ( as in the hip ) , while on the other hand s@xmath1 is neutralized by o@xmath2 ( as in the lip ) .
assume also that o@xmath2 is produced by the sequence ( r6)-(r11 ) , and that the o@xmath2 destruction rate increases if the o@xmath2 density decreases .
this is possible when reaction ( r12 ) between o@xmath2 and c is the dominant destruction mechanism for both species , so that @xmath87 increases if @xmath88 decreases . for these conditions
a chemical instability can occur , as we demonstrate with the following perturbation analysis .
consider a possible solution to the steady - state rate equations . with the above assumptions this requires @xmath89 .
now consider a perturbative increase in the ionization fraction @xmath90 , carried out at fixed gas density @xmath60 .
this will reduce h@xmath0 by a factor that depends on the relative efficiencies of dissociative recombination versus reaction with co in removing the h@xmath0 .
the o@xmath2 is removed by c ( via [ r12 ] ) , so for a given c density the reduction in h@xmath0 will lead to an equal reduction in o@xmath2 ( since o@xmath2 is formed by [ r6]-[r11 ] ) .
however , the decrease in o@xmath2 will lead to an increase in c , since the c is removed by o@xmath2 ( again via [ r12 ] ) .
this reduces o@xmath2 further still .
the s@xmath1 density then increases by a factor equal to the total reduction in o@xmath2 ( since s@xmath1 is removed by o@xmath2 ) .
finally , to close the loop , @xmath91 must then be set equal to the new value of @xmath92 , since by assumption s@xmath1 carries the positive charge . however , if the increase in s@xmath1 is larger than the initial increase in the electron density the cycle `` runs away '' in a positive feedback loop .
this h@xmath0-o@xmath2-s@xmath1 cycle is illustrated schematically in figure 2 .
it is clear that the cycle can also lead to a run - away decrease in the electron density , in response to a perturbative decrease in @xmath91 . for increasing electron densities , o@xmath2 decreases and c increases .
for decreasing electron densities , o@xmath2 increases and c decreases .
the instability shuts off for sufficiently large or small fractional ionizations . for large @xmath93 , radiative recombination , rather than removal by o@xmath2 ,
becomes the dominant s@xmath1 neutralization process . for small
@xmath93 , reaction with co replaces dissociative recombination as the h@xmath0 removal mechanism .
either way , the h@xmath0-o@xmath2-s@xmath1 instability cycle shuts off , and the gas becomes chemically stable . in the hip ,
dissociative recombination dominates the destruction of h@xmath0 , but s@xmath1 is removed by radiative recombination . in the lip ,
s@xmath1 is removed by o@xmath2 , but h@xmath0 is removed by co. crucially therefore , the h@xmath0o@xmath2s@xmath1 cycle is inoperative on the hip and lip branches , and these solutions are chemically stable .
we now identify the maximum possible density for the hip , @xmath51 , as the point at which o@xmath2 , _ as formed and destroyed in the hip _ , becomes sufficiently abundant to begin removing the s@xmath1 .
when this occurs , the h@xmath0o@xmath2s@xmath1 cycle turns on ( since h@xmath0 is still being removed by electrons ) and a stable hip is no longer possible . as the density is increased above @xmath51 the instability induces a sharp drop in @xmath91 . as the electron density drops in the run - away loop , o@xmath2 increases and c decreases .
the system is finally stabilized when the electron density becomes sufficiently small for reactions with co to begin dominating the removal of h@xmath0 .
the h@xmath0o@xmath2s@xmath1 cycle then turns off , and a stable solution on the lip is reached .
similarly , we identify the minimum possible density for the lip , @xmath53 , as the point at which the fractional ionization , _ as determined for the lip _ , becomes sufficiently large for dissociative recombination to begin dominating the removal of h@xmath0 .
when this occurs , the h@xmath0o@xmath2s@xmath1 cycle turns on , ( since s@xmath1 is still removed by o@xmath2 ) and a stable lip is no longer possible .
as the density is decreased below @xmath53 the instability induces a sharp rise in @xmath91 . as the electron density increases , o@xmath2 decreases and c increases .
the electron density rises until it becomes sufficiently large ( and the o@xmath2 density sufficiently small ) for radiative recombination to dominate the neutralization of s@xmath1 .
the instability then shuts off , and a stable solution on the hip is reached .
the mix branch can now also be understood .
it represents the locus of equilibrium solutions , with varying @xmath60 , for which the h@xmath0-o@xmath2-s@xmath1 cycle _ is _ operative , but for which any change in electron density is precisely matched by a corresponding change in the s@xmath1 density . for these solutions , dissociative recombination and reactions with co both
contribute to the removal of h@xmath0 , and s@xmath1 is removed by o@xmath2 . furthermore ,
reaction ( r12 ) is the dominant destruction mechanism , for both o@xmath2 and c. as the electron fraction increases along the mix , the o@xmath2 formation rate decreases , but by a factor that is smaller than the increase in electron density .
the density of c atoms ( which remove the o@xmath2 ) increases such that the total increase in o@xmath2 , and the resulting increase in s@xmath1 , exactly matches the increase in electron density .
figure 3 summarizes the key characteristics of the hip , lip , and mix branches , as described above .
the triangular symbols indicate the three `` legs '' of the h@xmath0-o@xmath2-s@xmath1 cycle illustrated in figure 2 . in the hip , removal of s@xmath1 by o@xmath2
is inoperative so this leg is labeled as `` off '' . in the lip ,
dissociative recombination of h@xmath0 is inoperative so this leg is `` off '' for this phase . for the mix , which shares properties of both the hip and lip , all three legs of the h@xmath0-o@xmath2-s@xmath1 cycle operate , and are `` on '' . in figure 3
we also indicate the dominant removal reactions for s@xmath1 , h@xmath0 , o@xmath2 , and c , for each of the three branches .
the hip and lip as originally defined by pineau des forets et al . (
1992 ) , were distinguished by whether h@xmath0 is removed by dissociative recombination ( hip ) , or by reactions with co ( lip ) .. however , they did not consider how the instability is quenched in each phase . ] here we have identified a second , and crucial , distinction . in the hip , atomic ions , are removed by ( slow ) radiative recombination . in the lip
the atomic ions are effectively removed by rapid dissociative recombination , via the formation of intermediate molecular ions , as in the sequence ( r22 ) and ( r23 ) . in our discussion
it is s@xmath1 that plays the crucial role of positive charge carrier in both the hip and lip .
the reason why simultaneous hip , lip , and mix solutions are possible now emerges .
bistability occurs when radiative recombination is able to dominate the removal of s@xmath1 in the hip up to maximal gas densities that are _ greater _ than the minimal densities down to which reactions with co dominate the removal of h@xmath0 in the lip .
alternatively , bistability occurs when it is possible for a low hip o@xmath2 abundance to be maintained up to maximal gas densities that are greater than the minimal gas densities down to which a large lip o@xmath2 density can be maintained .
this is our key result .
when this occurs , there exists a range of densities for which the s@xmath1 ions are either removed slowly via radiative recombination ( hip , low o@xmath2 ) or rapidly via dissociative recombination ( lip , high o@xmath2 ) .
the simultaneous solutions are separated by the h@xmath0-o@xmath2-s@xmath1 instability cycle , which drives the o@xmath2 density to either small ( hip ) or large ( lip ) values
. the critical densities @xmath53 and @xmath51 , as defined above , depend on a variety of factors including the elemental abundances , and several specific chemical rate coefficients . in
4 we derive simple analytic expressions for these critical densities .
we determine , analytically , when it is possible for @xmath51 to exceed @xmath53 , which is our condition for bistability , and we determine the range of gas densities for which bistability occurs .
for stable hip conditions the electron density must be sufficiently large for dissociative recombination to dominate the removal of h@xmath0 , and the o@xmath2 density must be sufficiently small for radiative recombination to dominate the removal of s@xmath1 .
we now show analytically that consistent solutions that satisfy these hip conditions are possible provided the density is less than a critical _ maximum _ gas density .
we first derive an expression for the o@xmath2 density in the low density limit of the hip .
we show that in this limit the o@xmath2 density is small , and radiative recombination dominates the removal of s@xmath1 .
we then show that our low density expression for o@xmath2 remains consistent only up to a maximal gas density .
we identify this as the maximum possible density for hip solutions .
we then verify that dissociative recombination dominates the removal of h@xmath0 , a necessary condition for the hip , for all densities below this critical density . in the hip ,
o@xmath2 is produced by the sequence ( r6)-(r11 ) , and is removed by reaction ( r12 ) with atomic carbon .
this gives , @xmath95 where @xmath96 and @xmath97 @xmath16 s@xmath17 are the rate coefficients ( at 50 k ) of ( r6 ) and ( r12 ) , and @xmath98 is the fraction ( jensen et al .
2000 ) of h@xmath48o@xmath1 recombinations that produce oh . atomic carbon is produced by helium impact ionization of co followed by immediate charge transfer neutralization ( [ r15 ] and [ r16 ] ) , and is removed by cosmic - ray induced photoionization ( [ r18 ] ) . because ( r15 ) dominates the removal of he@xmath1 , the rate of c formation equals the cosmic - ray ionization rate of helium .
therefore , @xmath99 independent of @xmath100 . in this expression @xmath101 is the efficiency of ( r18 ) , and @xmath102 is the helium abundance .
it is clear that a large c density is indeed expected in the hip , since @xmath103 is comparable to the typical total gas - phase carbon abundance ( e.g. @xmath104 in our numerical model ) . the large c density keeps the o@xmath2 abundance small in the hip .
the o@xmath2 density is proportional to h@xmath0 , which in turn , depends on the electron density . for electrons we set @xmath89 .
the s@xmath1 is produced the same way as atomic carbon , i.e. , by helium impact ionization of co followed by charge transfer ( [ r15 ] and [ r16 ] ) .
we assume that the s@xmath1 is removed by radiative recombination .
this gives , @xmath105 where the radiative recombination rate coefficient @xmath106 @xmath16 s@xmath17 . in this limit
the fractional ionization varies as @xmath107 . for @xmath108 @xmath39 , equation ( [ efrac1 ] ) gives @xmath109 in good agreement with @xmath110 computed numerically ( see figure 1 ) .
h@xmath0 is produced by cosmic - ray ionization and is removed by dissociative recombination , so that @xmath111 where @xmath112 @xmath16 s@xmath17 is the total rate coefficient for dissociative recombination .
the h@xmath0 abundance also varies as @xmath107 . for @xmath108 @xmath39 , equation ( [ h3plus1 ] ) gives @xmath113 , as also computed numerically .
inserting equations ( [ c1 ] ) , ( [ efrac1 ] ) , and ( [ h3plus1 ] ) into equation ( [ oxy1 ] ) gives @xmath114 where @xmath115 , and where we have assumed that most of the oxygen not bound in co remains atomic .
the o@xmath2 abundance varies as @xmath107 . for @xmath108 @xmath39 , equation ( [ oxy2 ] ) gives @xmath116 , close to @xmath117 computed numerically .
the o@xmath2 density as given by equation ( [ oxy2 ] ) , is sufficiently small for removal of s@xmath1 by o@xmath2 to be negligible compared to radiative recombination .
this can be seen by considering the ratio of the s@xmath1 removal rates @xmath118 where @xmath119 @xmath16 s@xmath17 is the rate coefficient for reaction ( r22 ) . because this ratio is independent of @xmath100 ,
the removal of s@xmath1 by o@xmath2 is negligible at all densities so long as equation ( [ oxy2 ] ) for the o@xmath2 abundance is an appropriate approximation .
equation ( [ oxy2 ] ) is valid so long as cosmic - ray photons dominate the removal of the carbon atoms , so that the c density is given by equation ( [ c1 ] ) .
however , cosmic - ray induced destruction occurs at a rate independent of @xmath60 , whereas the o@xmath2 density as given by equation ( [ oxy2 ] ) varies as @xmath120 .
therefore , for sufficiently large @xmath60 reactions with o@xmath2 ( [ r12 ] ) must eventually dominate the removal of the c atoms .
when this occurs the c density becomes small , and the o@xmath2 density must become large ( see equation [ [ oxy1 ] ] ) . we identify this as the point where o@xmath2 becomes sufficiently abundant to begin removing the s@xmath1 .
however , because h@xmath0 continues to be removed by dissociative recombination , the h@xmath0-o@xmath2-s@xmath1 instability cycle turns on at this point , and a stable hip is no longer possible .
thus , the hip can be maintained so long as the o@xmath2 density given by equation ( [ oxy2 ] ) is small enough for cosmic - ray photoionization to dominate the removal of the c atoms .
this condition , @xmath121 provides a maximal density for hip solutions . for our standard elemental abundances and rate - coefficients
@xmath122 this is slightly larger than our numerically computed density of 675 @xmath39 at which the transition to the lip occurs ( see figure 1 ) . for consistency we must verify that dissociative recombination dominates the removal of h@xmath0 for densities as high as @xmath123 .
dissociative recombination dominates when the ratio @xmath124 where @xmath125 @xmath16 s@xmath17 is the rate coefficient of reaction ( r5 ) . for @xmath126 @xmath39
this ratio equals @xmath127 , so consistency is obtained . in the lip , the electron density must be sufficiently small for co to dominate the removal of h@xmath0 , and the o@xmath2 density must be sufficiently large for o@xmath2 to dominate the removal of s@xmath1 .
we now show that these conditions can be satisfied , provided the gas density is greater than a critical _ minimum _ density .
to show this , we first derive an expression for the electron density assuming that s@xmath1 is removed by abundant o@xmath2 , with an upper limit for the o@xmath2 density set by cosmic - ray induced photodissociation .
we show that co then dominates the removal of h@xmath0 for densities greater than a minimum critical density .
we then verify that the assumption that o@xmath2 dominates the removal of s@xmath1 is consistent for densities greater than this critical density .
for the electron density in the lip we again assume that @xmath89 . in the lip , s@xmath1 is produced by cosmic - ray induced photoionization , and we assume that it is removed by o@xmath2 .
this gives , @xmath129 where @xmath130 is the efficiency of ( r21 ) .
h@xmath0 is removed by co , so that @xmath131 where we have assumed that in the lip all of the carbon is locked in co. for @xmath132 , equation ( [ le1 ] ) gives @xmath133 , in good agreement with the numerically computed @xmath134 .
for o@xmath2 we again assume the production sequence ( r6)-(r11 ) .
if the o@xmath2 is removed by only cosmic - ray induced photodissociation , @xmath135 where @xmath136 is the efficiency of ( r13 ) .
this is actually an upper - limit for the o@xmath2 abundance because additional chemical reactions contribute to the removal of the o@xmath2 in the lip . in this approximation
the o@xmath2 density is independent of @xmath100 , consistent with our numerical computations ( see figure 1 ) .
because the upper limit is close to the total available oxygen not locked in co , we set @xmath137 , where the o@xmath2 fraction @xmath138 . in our numerical computation
@xmath139 ( see figure 1 ) .
the electron fraction , @xmath93 , is then @xmath140 in this limit @xmath93 varies as @xmath141 . for @xmath142 , equation ( [ le4 ] )
gives @xmath143 , in good agreement with @xmath144 in figure 1 .
the electron density as given by equation ( [ le4 ] ) is independent of @xmath60 , whereas the co density decreases with @xmath60 .
reactions with co can therefore dominate the removal of h@xmath0 only down to a minimal density set by the condition @xmath145 this relation provides a minimal critical density @xmath146 for which lip solutions are possible . for @xmath142 ,
this is slightly smaller than the numerically computed value of 370 @xmath39 at which the transition from the lip to hip occurs ( see figure 1 ) .
when the gas density falls below the critical density electrons remove h@xmath0 , but o@xmath2 continue to remove s@xmath1 .
the h@xmath0-o@xmath2-s@xmath1 instability cycle therefore turns on , and the gas is driven from the lip to the hip . for consistency we must now verify that reactions with o@xmath2 dominate the removal of s@xmath1 everywhere in the lip
this requires that @xmath147 for all lip densities , down to the minimal density .
inserting @xmath148 as given by equation ( [ le6 ] ) gives @xmath149 .
we conclude that consistent lip solutions are obtained even for @xmath150 as small as @xmath151 ( for which the o@xmath2 density in the lip is still much larger than in the hip . )
the key result of the above analysis is that for our standard parameters @xmath152 , as given by equation ( [ hcond2 ] ) , is _ greater _ than @xmath148 , as given by equation ( [ le6 ] ) .
the gas is therefore bistable between the maximal hip density and minimum lip density .
the mix branch must therefore `` switch back '' from the maximal hip density to the minimal lip density .
the h@xmath0-o@xmath2-s@xmath1 instability cycle drives the o@xmath2 density to either very low values in the hip where a large ionization fraction is maintained by slow radiative recombination , or it drives the o@xmath2 density to very large values in the lip where a low ionization fraction is maintained by rapid dissociative recombination .
our analytic estimates for the two critical densities are in good agreement with the numerical results displayed in figure 1 .
equations ( [ hcond ] ) and ( [ le5 ] ) show that the critical densities depend on several parameters , including the gas - phase elemental abundances of carbon , oxygen , sulfur and helium , and on several specific chemical rate coefficients , including @xmath153 , the dissociative recombination rate coefficient of h@xmath0 .
varying any of these parameters alters the range of densities for which bistability occurs , or eliminates it entirely if @xmath152 becomes smaller than @xmath148 . a dependence of bistability on @xmath153 , a rate coefficient that has been the subject of much investigation and controversy ( see e.g. , mccall et al .
2004 ) , was found in numerical computations by pineau des forets & roueff ( 2000 ) . in their calculations
( see their fig .
4 ) they set @xmath154 @xmath39 , and varied @xmath153 from @xmath155 to @xmath156 @xmath16 s@xmath17 .
they found that the gas is in the lip or hip , for low or high values of @xmath153 , and that it is bistable for intermediate h@xmath0 recombination efficiencies .
our analysis provides an explanation for this behavior . in figure 4
we plot @xmath93 versus @xmath38 , for five values of @xmath153 , ranging from 0.2 to 5 times our standard 50 k value of @xmath157 @xmath16 s@xmath17 . in these numerical computations , all other parameters , including all rate coefficients and gas - phase elemental abundances , are the same as in the computations we presented in 2 .
figure 4 shows that for small @xmath153 bistability is suppressed , and that as @xmath153 is increased the gas becomes bistable for an increasing range of densities
. for @xmath38 held fixed at , say , @xmath158 @xmath39 , the gas is in the lip or hip for low or high @xmath153 , and is bistable for intermediate values .
this is the same behavior found by pineau des forets & roueff ( 2000 ) .
these results can be readily understood as follows . as the dissociative recombination rate coefficient of h@xmath0 is increased , less o@xmath2 is produced by the sequence ( r6)-(r11 ) at any give density .
an hip can therefore be maintained to higher densities before o@xmath2 becomes significant in removing s@xmath1 .
an increased h@xmath0 recombination efficiency also implies that electrons become competitive with co in removing h@xmath0 at larger densities in the lip .
quantitatively , equation ( [ hcond ] ) shows that @xmath51 , the density above which o@xmath2 begins removing s@xmath1 , is proportional to @xmath159 .
however , @xmath53 , the density below which electrons begin removing the h@xmath0 , varies only linearly with @xmath153 .
therefore , as @xmath153 is reduced , @xmath152 falls _ below _
@xmath53 , and bistability disappears .
as @xmath153 is increased , @xmath51 grows more rapidly , and eventually exceeds , @xmath53 , and the gas becomes bistable for an increasing range of densities .
for example , if @xmath153 is increased by a factor of 5 relative to our standard value , figure 4 shows that @xmath51 increases by about a factor of 25 , and @xmath53 increases by about a factor of 5 . this is consistent with expressions ( [ hcond ] ) and ( [ le5 ] ) .
a similar scaling analysis may be carried out for the other parameters that control @xmath160 and @xmath123 .
for example , if the oxygen abundance is reduced our analysis shows that the regime of bistability should move to higher values of @xmath100 , as found numerically by viti et al .
( 2001 ) ( see their figures 5 and 7 ) . for a lower oxygen abundance @xmath160 is increased because the hip can be maintained to a higher density before o@xmath2 becomes abundant enough to remove the atomic ions .
the reduced o@xmath2 density leads to a higher electron density in the lip , so @xmath160 is increased as well .
our numerical and analytic results show how the onset of the h@xmath0-o@xmath2-s@xmath1 instability cycle depends on the o@xmath2 and electron densities as determined by the elemental abundances and by the h@xmath0 dissociative recombination rate coefficient .
however , figure 4 reveals an additional interesting feature . for @xmath153
equal to 0.2 times our standard value the gas again becomes marginally bistable at @xmath161 @xmath39 .
at such low densities the dominant atomic ion in the hip is c@xmath1 rather than s@xmath1 ( see fig . 1 ) .
the instability , and the transition from hip to lip , is therefore modified by the addition of an identical h@xmath0-o@xmath2-c@xmath1 cycle , with reactions ( r19 ) and ( r20 ) operating to neutralize the c@xmath1 .
the behavior in this numerical example is a bit more complicated because s@xmath1 is still the dominant ion in the lip at the transition density .
we note that at such low densities realistic clouds would likely become `` diffuse '' and optically thin to external far - ultraviolet radiation , with a resulting reduction in o@xmath2 and further modifications to the chemistry .
we do not consider such effects here .
we have shown that bistability arises in purely gas - phase models , consistent with previous findings . when the hip and lip solutions `` overlap ''
, neutralization of atomic ions ( s@xmath1 in our models ) proceeds by either slow radiative recombination in the hip , or via the formation of intermediate molecular ions ( @xmath3 ) followed by rapid dissociative recombination in the lip .
the two solutions are separated by a chemical instability , which we have identified in our models as the h@xmath0-o@xmath2-s@xmath1 cycle .
our analysis suggests that bistability will be damped , or eliminated entirely , if additional neutralization processes are included that interrupt the h@xmath0-o@xmath2-s@xmath1 cycle .
we demonstrate this by considering how the single process of `` grain assisted recombination '' of atomic ions affects the chemistry . in this process ,
atomic ions recombine via electron transfer from grains to the ions during collisions with the grains ( draine & sutin 1987 ; weingartner & draine 2001 ) . the grain assisted recombination rate may be expressed as @xmath162 , where @xmath26 is the total density of hydrogen nuclei . for any ion
, @xmath163 is an effective rate coefficient that depends on the dust - to - gas ratio , grain size distribution , grain charge , and gas temperature .
this process dominates radiative recombination when @xmath164 , where @xmath165 is the radiative recombination rate coefficient . for @xmath166 @xmath16 s@xmath17 ( appropriate for s@xmath1 ) , and a fractional ionization @xmath167 , as occurs for densities near @xmath51 in our computation ( see fig . 1 ) , grain assisted recombination dominates radiative recombination for @xmath168 @xmath16 s@xmath17 .
this is well below the maximum values of @xmath169 @xmath16 s@xmath17 found by weingartner & draine , so this process is potentially relevant in determining the ionization balance .
furthermore , grain assisted recombination will interfere with the h@xmath0-o@xmath2-s@xmath1 cycle and render it inoperative when @xmath170 , where as previously defined , @xmath171 is the fraction of oxygen bound in o@xmath2 , and @xmath119 @xmath16 s@xmath17 is the rate coefficient of reaction ( r22 ) . for @xmath139 and @xmath172 ,
the h@xmath0-o@xmath2-s@xmath1 instability cycle is quenched when @xmath173 @xmath16 s@xmath17 .
bistability and the sharp transition from hip to lip should then disappear .
we illustrate this in figure 5 , where we plot @xmath93 versus @xmath38 , for @xmath163 ranging from zero ( as in figures 1 and 4 ) to @xmath174 @xmath16 s@xmath17 , for all atomic ions .
all other parameters are as in the calculation in 2 . as @xmath163
is increased the diminished electron density leads to more h@xmath0 and o@xmath2 in the hip , so that @xmath51 is reduced .
similarly , the onset of dissociative recombination in the lip occurs at lower densities , and @xmath53 is reduced .
thus , at first the bistability region shifts to lower densities as @xmath163 is increased .
however , as estimated above , when @xmath163 exceeds @xmath175 @xmath16 s@xmath17 bistability is eliminated entirely , because the h@xmath0-o@xmath2-s@xmath1 ( and h@xmath0-o@xmath2-c@xmath1 ) cycles are fully quenched , and neutralization occurs entirely by grain recombination .
these results strongly suggest that for realistic clouds , in which gas - grain neutralization is a likely important process , bistability will not occur .
the precise value of @xmath163 in dark clouds may be variable , as the grain population can be modified by coagulation ( draine 1985 ; flower et al .
2005 ) and other processes . however , even with the removal of all grains with sizes below @xmath176 , the rate coefficient @xmath177 @xmath16 s@xmath17 ( e.g. mckee 1989 ) , which is sufficiently large to suppress bistability .
we thus confirm and explain the results of shalabiea and greenberg ( 1995 ) , who found that bistability is eliminated when gas - grain processes , including grain assisted recombination , are incorporated in the numerical computations .
the non - linearity of equations ( 1)-(4 ) is a mathematical prerequisite for bistability and the appearance of multiple solutions ( le bourlot et al . 1993 , 1995 ;
lee et al .
1998 ; pineau des forets & roueff 2000 ; charnley & markwick 2003 ) .
however , the purpose of our paper has been to identify and analyze the chemical mechanisms underlying the bistability phenomenon .
we have shown that a simple chemical instability , involving atomic ions ( typically s@xmath1 ) , o@xmath2 molecules , and h@xmath0 ions , allows multiple solutions to appear near the transition points from low - density high - ionization ( hip ) to high - density low - ionization ( lip ) conditions in the gas - phase chemistry .
the instability drives the o@xmath2 density to either small ( hip ) or large ( lip ) values .
bistability occurs when a low o@xmath2 abundance in the hip can be maintained up to maximal densities that are greater than the minimal densities above which a large o@xmath2 density is maintained in the lip .
two modes of neutralization are then simultaneously available to the gas .
the slow ( hip ) mode is radiative recombination ( usually @xmath178 ) .
the fast ( lip ) mode is formation of molecular ions followed by dissociative recombination ( usually @xmath3 ; @xmath179 ) .
we have derived simple analytic scaling relations for the range of gas densities and ionization rates for which bistability occurs .
our formulae show how the bistable range depends on the assumed gas - phase abundances , and several reaction rate - coefficients , including the dissociative recombination rate coefficient of h@xmath0 .
bistability is eliminated when the two modes of gas - phase neutralization are superseded by gas - grain processes such as grain assisted recombination of the atomic ions .
this likely happens for the range of electron and o@xmath2 densities where bistability typically occurs in purely gas - phase models .
we conclude that the bistability phenomenon probably does not occur in realistic dusty interstellar clouds , although it remains of theoretical interest .
we thank a. dalgarno , o. gnat , c.f .
mckee , and d. neufeld for discussions .
we thank a. markwick - kemper for constructive comments on our manuscript .
this research is supported by the israel science foundation , grant 221/03 .
cardelli , j.a . , mathis , j.s . ,
ebbets , d.c . , & savage , b.d.1993 , , 402 , l17 cardelli , j.a . ,
sofia , u.j . ,
savage , b.d . ,
keenan , f.p .
, & dufton , p.l.1994 , , 420 , l29 charnley , s.b . , & markwick , a.j .
2003 , , 399 , 583 draine , b.t .
1985 , in protostars and planets ii , ed .
d. black and m. matthews ( tucson : university of arizona press ) , p. 33 .
le bourlot , j. , pineau des forets , g. , roueff , e. , & schilke , p.1993 , apj , 416 , l87 le bourlot , j. , pineau des forets , g. , roueff , e. , & flower , d. r.1995 , , 302 , 870 le bourlot , j. , pineau des forets , g. , & roueff , e. 1995 , , 297 , 251 lee , h .- h . ,
roueff , e. , pineau des forets , g. , shalabiea , o. m. , terzieva , r. , & herbst , eric 1998 , , 334 , 1047 lepp , s. , dalgarno , a. , van dishoeck , e.f . , & black , j.h.1988 , , 329 , 418 lepp , s. , & dalgarno , a. 1996 , , 306 , l21 le teuff , y.h . ,
millar , t.j .
, & markwick , a.j2000 , , 146 , 157 1996 , , 466 , 561 mccall , b.j .
2004 , phys.rev.a , 70 , e2716 o@xmath2s@xmath1 cycle . increasing the electron density ,
reduces h@xmath0 via dissociative recombination .
this reduces o@xmath2 formed by the sequence ( r6)-(r11 ) .
destruction of s@xmath1 by o@xmath2 decreases , so the s@xmath1 density increases .
this cycle can `` run away '' when s@xmath1 is the dominant positive charge carrier ( see text ) .
, width=497,height=718 ] lllllllll h & & 1.7(-1 ) & & 1.7(-1 ) & & 1.7(-1 ) & & 1.0 + h@xmath2 & & 4.3(2 ) & & 4.3(2 ) & & 4.3(2 ) & & 1.0 + c & & 1.1(-2 ) & & 1.3(-3 ) & & 2.2(-4 ) & & 50.0 + n & & 2.8(-2 ) & & 1.9(-2 ) & & 1.0(-2 ) & & 2.8 + o & & 1.4(-1 ) & & 1.2(-1 ) & & 1.0(-1 ) & & 1.4 + s & & 2.6(-3 ) & & 3.8(-3 ) & & 3.6(-3 ) & & 0.72 + si & & 2.1(-5 ) & & 2.9(-5 ) & & 2.7(-5 ) & & 0.78 + e & & 4.4(-3 ) & & 1.7(-3 ) & & 5.3(-4 ) & & 8.30 + c@xmath1 & & 9.0(-5 ) & & 3.1(-5 ) & & 1.5(-5 ) & & 6.0 + n@xmath1 & & 5.0(-10 ) & & 5.4(-10 ) & & 6.0(-10 ) & & 0.83 + o@xmath1 & & 4.2(-11 ) & & 1.1(-10 ) & & 1.9(-10 ) & & 0.22 + s@xmath1 & & 3.3(-3 ) & & 1.3(-3 ) & & 4.3(-4 ) & & 7.67 + si@xmath1 & & 9.8(-4 ) & & 2.7(-4 ) & & 6.1(-5 ) & & 16.1 + h@xmath0 & & 3.6(-6 ) & & 6.3(-6 ) & & 9.0(-6 ) & & 0.40 + h@xmath48o@xmath1 & & 9.6(-8 ) & & 4.4(-7 ) & & 1.9(-6 ) & & 0.051 + hco@xmath1 & & 2.3(-7 ) & & 1.2(-6 ) & & 5.2(-6 ) & & 0.044 + hcs@xmath1 & & 7.3(-8 ) & & 2.1(-7 ) & & 9.6(-7 ) & & 0.076 + co@xmath1 & & 2.2(-11 ) & & 5.8(-11 ) & & 9.5(-11 ) & & 0.23 + no@xmath1 & & 1.1(-8 ) & & 4.6(-8 ) & & 1.2(-7 ) & & 0.092 + o@xmath180 & & 2.5(-9 ) & & 2.8(-8 ) & & 1.9(-7 ) & & 0.013 + so@xmath1 & & 3.7(-8 ) & & 1.6(-7 ) & & 5.2(-7 ) & & 0.071 + ch & & 2.7(-6 ) & & 7.3(-7 ) & & 3.0(-7 ) & & 9.0 + nh & & 1.1(-6 ) & & 1.5(-6 ) & & 2.0(-6 ) & & 0.55 + oh & & 2.8(-5 ) & & 6.3(-5 ) & & 1.1(-4 ) & & 0.25 + h@xmath2o & & 2.8(-4 ) & & 1.1(-3 ) & & 2.1(-3 ) & & 0.13 + sh & & 3.9(-7 ) & & 6.1(-7 ) & & 1.0(-6 ) & & 0.39 + sih & & 1.6(-7 ) & & 5.0(-8 ) & & 1.2(-8 ) & & 13.3 + co & & 1.0(-1 ) & & 1.1(-1 ) & & 1.1(-1 ) & & 0.91 + no & & 3.0(-5 ) & & 1.2(-4 ) & & 3.0(-4 ) & & 0.1 + o@xmath2 & & 4.2(-4 ) & & 3.6(-3 ) & & 1.2(-2 ) & & 0.035 + so & & 6.5(-6 ) & & 1.2(-4 ) & & 8.0(-4 ) & & 0.008 + sio & & 5.3(-4 ) & & 1.2(-3 ) & & 1.4(-3 ) & & 0.38 + cs & & 1.2(-3 ) & & 1.8(-3 ) & & 2.2(-3 ) & & 0.55 + cn & & 4.5(-6 ) & & 2.0(-6 ) & & 1.3(-6 ) & & 3.46 + hcn & & 4.6(-5 ) & & 1.0(-5 ) & & 3.2(-6 ) & & 14.4 + n@xmath2 & & 1.8(-2 ) & & 2.2(-2 ) & & 2.7(-2 ) & & 0.67 + | we present an analysis of `` bistability '' in gas - phase chemical models of dark interstellar clouds .
we identify the chemical mechanisms that allow high- and low - ionization solutions to the chemical rate - equations to coexist .
we derive simple analytic scaling relations for the gas densities and ionization rates for which the chemistry becomes bistable .
we explain why bistability is sensitive to the h@xmath0 dissociative recombination rate coefficient , and why it is damped by gas - grain neutralization . |
the key scheme of cavity quantum electrodynamics ( qed)@xcite , modern atom optics and interferometry @xcite is the resonant interaction of an atom with cavity fields , created as standing or counterpropagating waves . the first type ( with respective quantization on the sw basis ) is attained in two plane - parallel - mirror cavities ( in short , cavity ) , while the second type ( with respective cpw - quantization ) is attained in three or more mirror cavities ( ring cavities ) .
it is well known , in addition , that for classical picture of fields , these two representations are equivalent in the sense , that the sw can always be presented as a superposition of two counterpropagating travelling waves . in the quantum theory , nevertheless , they are divers@xcite , including the hamiltonians of interaction . hence it needs to be ascertained : are the diversities only quantitative or qualitative too , and how much are they for these or that circumstances and processes .
note , that it presents not only an academic interest , since in the microcavities ( @xmath0 ) several dozens of photons can induce strong optical nonlinearities for dipole - allowed optical transitions .
as far as we know , elucidation of physical aspects of mentioned incongruity had been done in paper @xcite , taking as an example the process of near - resonant coherent diffraction of atomic matter - waves by a space - periodic laser radiation field , known as near - resonant kapitza - dirac diffraction@xcite . by means of numerical solution of the master equations for atomic probability amplitudes ,
the existence of evident differences between sw and cpw diffraction patterns was shown , in strictly quantum domain of cavity fields ( the mean number of photons in cavities was chosen one ) .
the physical reason of the difference also was presented .
it is the behavior of momentum conservation law in relevant atom + field systems .
this general principle ceases to apply in sw - case , because the atom changes its own momentum due to the interaction , while the sw - field does not . in contrary , in cpw - case
both , atomic and field subsystems , change the momenta equally and in opposite directions , thereby conserving the systems total momentum . already on the base of this sole difference
can be anticipated a
diffraction for cpw - case with a respectively narrower distribution of states in the momentum space . in this paper are considered the stationary states of the system two - level atom + quantized field in both sw and cpw - cases and the qualitative and quantitative differences between them are carried out in the range of quantum optics ( the number of photons should be less than several tens ) . to this end the simplest theoretical model , the jaynes - cummings model ( jcm ) @xcite , including the quantized atomic center - of - mass motion @xcite , will be used .
the spontaneous emission and other incoherent processes , as usual for this model , are omitted . to make possible
the direct comparison of these two concepts we , at first , represent detailed theory for stationary states of the systems at hand .
our representation for sw - case somewhat coincides with the picture in @xcite .
the preliminary analysis of the states and the comparison of the sw and cpw - cases are based on numerical solutions of the exact , as well as of the off - resonant approximate equations of atomic probability amplitudes in momentum space .
an additional approximate analytical solution of the problem allows us to find out the parameters ( terms ) , determining the size of nonequivalency between the sw and cpw - spectrums and eigenfunctions in explicit form .
the latter solutions pertain , however , only to bounded ( wells inner ) atomic states .
the paper is organized as follows .
section ii represents the basic model and the exact set of equations .
here we also give some numerical simulations of the problem , implemented for the sodium atom with @xmath1 main transition .
section iii represents the off - resonance and mentioned analytical approximations . in section
iv we examine the bragg and doppleron resonances , including the recoil energy terms . in section v we consider the mechanisms responsible for the formation of stationary states from free definite - momentum ones . conclusions and some short remarks about the subject
are given in section vi .
we consider a two level atom of mass @xmath2 and optical transition frequency @xmath3 , interacting with a quantized plane monochromatic field of frequency @xmath4 . we will restrict ourselves with two important cases of standing and counterpropagating waves .
the longitudinal components @xmath5 of the atomic center - of - mass ( c.m . )
momentum remain unchanged for perfectly plane fields , so that only the transverse atomic momentum @xmath6 needs to be considered .
since the incoherent processes are not included , the system is governed by the schrodinger equation .
let us first consider the cpw - case .
the hamiltonian in dipole and rotating - wave approximations is given by @xcite @xmath7 @xmath8 , \eqnum{1 } \label{1}\ ] ] where @xmath9 is atomic momentum operator ( 1d ) , @xmath10 and @xmath11 are usual pseudospin atomic ( pauli ) operators , @xmath12 and @xmath13 @xmath14 are annihilation and creation operators for two running wave modes , @xmath15 .
the factor @xmath16 in coupling constant is the vacuum rabi frequency , and is connected with the atomic transition dipole moment @xmath17 by relation @xmath18 , where @xmath19 is the normalization ( cavity ) volume .
the coherent binding between the internal and c.m .
variables is by the rabi frequency @xmath16 , too .
the system under consideration has four degrees of freedom : two per atomic internal and 1d c.m .
motions , and two per counterpropagating travelling waves .
three operators , forming together with the hamiltonian the complete set of mutually commutative operators , are known and are the following : the excitation number operator @xmath20 the total momentum operator @xmath21 and the operator @xmath22 which combines translational @xmath23 and dipole inversing @xmath24 operators . therefore @xmath25 and @xmath26 have compatible eigenvalues and a common system of eigenfunctions .
only these states , being the basic , should be considered thereafter in this paper .
any other state , of course in principle , can be presented via these eigenstates .
note , that sometimes the operator @xmath27 is used in related problems .
however , @xmath28and can not be added into the set of mentioned operators or replace the operator @xmath29 there . out of a desire to better understand the future solutions and their connection with free - state values , which present a definite interest too
, it may be worth to emphasize here that all three @xmath30 and @xmath31 operators do nt contain the interaction parameter @xmath16 .
they are interaction independent .
their eigenvalues , denoted by @xmath32 and @xmath33 respectively , also are interaction independent .
therefore , any intermediate or final state , created from the initial one due to interaction should possess the same values of @xmath32 and @xmath33 as the free initial state .
these ones , of course , do nt need to be stationary in general , but may be such , if the interaction is adiabatic@xcite .
a sufficient condition for adiabatic following is the exceeding of the interaction switching on and switching off times ( @xmath34 ) , the inverse quantity of the system s smallest characteristic frequency ( @xmath35 ) . in the system at hand ,
as a such frequency appears the one - photon recoil frequency @xmath36 so for the interaction times essentially exceeding the recoil time @xmath37 , the atomic state evaluation is adiabatic with respect to both , internal and c.m . ,
motions of the atom @xcite .
failure of the adiabatic condition implies , respectively , that the population leaks into the immediate neighboring states , first of all into the translational states .
applying the general remarks to the system under consideration , we arrive to the possibility to treat the stationary states as attained adiabatically from some free initial state and then replace the set of @xmath32 and @xmath33 quantities by a new set of three related parameters of this initial state : atomic momentum @xmath38 and the photon numbers @xmath39and @xmath40 in counterpropagating waves
. one must keep in mind , however , that the mentioned replacement , which will be embodied soon , is substantial for adiabatically treated stationary states and is not obligatory even in this case .
dirac notations @xmath41and @xmath42are used for quantized travelling - wave states ( number states ) .
atomic internal ground and excited states are specified by @xmath43 and @xmath44 matrixes respectively , the c.m .
states- by the function @xmath45 hilbert space of the atom+field
system is the direct sum of hilbert spaces of atomic c.m .
and internal motions , as well as of the field modes .
thereby , the nondegenerate wave function of the free system can be written as @xmath46 where @xmath47 is the normalization constant , @xmath48 @xmath49 and @xmath50 in general .
the seeking wave functions of the interacting system can be written as a superposition of ( [ 5])-type terms .
the general form of a wave function with a definite @xmath51 ( @xmath52 ) is @xmath53 @xmath54 where @xmath55 and @xmath56 are arbitrary coefficients yet .
admissible solutions also should be normalized according to @xmath57 @xmath58 note , that the excitation number @xmath51 can be represented by means of the numbers @xmath39 and @xmath59 ( see the remark above ):
@xmath60 validity of which can be directly checked with the help of wave function ( [ 6 ] ) . for the next step we demand from @xmath61 to be the eigenfunction of total momentum operator @xmath62 : @xmath63 inserting ( [ 3 ] ) and ( [ 6 ] ) into ( [ 8 ] ) we arrive to equations @xmath64
@xmath65 where @xmath66 is the dirac delta - function and @xmath67 and @xmath68 are the new coefficients .
the eigenvalue @xmath69 can be written as : @xmath70 where @xmath38 is chosen the same for both , ground and excited atomic levels .
integration over @xmath71 now yields to the following form : @xmath72 this form is already eigen - one for the operator @xmath29 @xmath73 @xmath74 hence , the expression ( [ 11 ] ) is the eigenfunction of operators @xmath75 and @xmath29 .
using([7 ] ) , ( [ 10 ] ) and the second equation in ( [ 13 ] ) , the latter three can be replaced by the initial values @xmath76 and @xmath77 as it was previously stated , we are concentrating on the stationary states , where @xmath78 @xmath79 being energy of the system .
after standard transformations the schrodinger equation leads to following set of tridiagonal recurrence algebraic equations : @xmath80 a_{r}=\left ( \hbar \omega _ { 0}/2^{3/2}\right ) \left [ \sqrt{r}b_{r-1}+% \sqrt{n - r}b_{r}\right ] , \eqnum{15a } \label{15a}\ ] ] @xmath81 b_{r}=\left ( \hbar \omega _ { 0}/2^{3/2}\right ) % \left [ \sqrt{r+1}a_{r+1}+\sqrt{n - r}a_{r}\right ] , \eqnum{15b } \label{15b}\ ] ] where @xmath82 and @xmath83 is the atom - field frequency detuning .
equations ( [ 15a ] ) , ( [ 15b ] ) , as well as equations ( [ 18a ] ) and ( [ 18b ] ) for the sw - case , are the basis of the remainder of this paper .
they determine the coefficients ( probability amplitudes ) @xmath55 , @xmath56 for ground and excited internal energy levels and the permitted values of energy @xmath84 the set of eqs .
( [ 15a ] ) , ( [ 15b ] ) has @xmath85 solutions
@xmath86 @xmath87 @xmath88 of them may be interpreted as created adiabatically from initial states with a ground - level atom , and the other @xmath51 , as created respectively with an excited - state atom . for simplicity
we suppress @xmath32 and @xmath89 in probability amplitude and energy notations . rather than pursuing the governing equations in this cpw - case we now turn to another case , to sw - case .
the respective hamiltonian is @xcite @xmath90 , \eqnum{16 } \label{16}\ ] ] where @xmath91 and @xmath92 are now sw - mode annihilation and creation operators , obeying the same - type commutation relation @xmath93=1 $ ] as the running modes .
note , that the coupling constant in this case is smaller by factor @xmath94 than in the former case .
the system
atom + sw has three degrees of freedom , by one less than the system with cpw .
the reason is , that the momentum @xmath95and is not commutative with the hamiltonian ( [ 16 ] ) .
proceeding the same way as for ( [ 11 ] ) and ( [ 15a ] ) , ( [ 15b ] ) , we arrive to @xmath96 for wave function , and @xmath97 a_{l}=\left ( \hbar \omega _ { 0}/4\right ) \sqrt{n}\left [ b_{l-1}+b_{l}\right ] , \eqnum{18a } \label{18a}\ ] ] @xmath98 b_{l}=\left ( \hbar \omega _ { 0}/4\right ) \sqrt{n}% \left [ a_{l+1}+a_{l}\right ] \eqnum{18b } \label{18b}\ ] ] for probability amplitudes and energies . here
@xmath99 , and the other notations are obvious .
for adiabatically attained states @xmath100 besides the apparent similarities , the basic sets of equations , ( [ 15a ] ) , ( [ 15b ] ) and ( [ 18a ] ) , ( [ 18b ] ) , also have basic differences .
while the first set is finite - size , the second one is infinite ( even for @xmath101 one - photon case ) .
the next difference is in right - hand side coefficients .
they are varied ( @xmath102-dependent ) for the first , but are constant for the second set . and
the final difference , which already is conceptual and deserves a peculiar attention , is that the total momentum @xmath69 in cpw - case , in contrast to its counterpart @xmath38 in sw case , can not be treated as a quasimomentum .
this assertion immediately follows from the fact , that the replacing @xmath103 , with the relabeling @xmath104 in ( [ 15a ] ) , ( [ 15b ] ) , @xmath105 being an integer , conserves the left - hand side coefficients invariant , while the right - hand side coefficients get new values , leading thereby to other values for eigenenergies and probability amplitudes .
reminding that the physical reason of this difference is the participation of the momentum conservation law , we arrive at a conclusion that @xmath106 @xmath107 @xmath108 @xmath106 @xmath109 @xmath110 @xmath111 @xmath112 @xmath113 @xmath114 @xmath115 @xmath116 @xmath106 @xmath117 @xmath108 @xmath118 @xmath119 .
@xmath120 @xmath114 @xmath121 @xmath122 @xmath108 @xmath123 @xmath111 @xmath124 @xmath125 @xmath126 @xmath127 @xmath128 @xmath129 @xmath130 the main goal of the remainder of this section will be to illustrate the behavior of both systems via the results of numerical calculations . while the solutions of the cpw equations have been got exactly , to sw equations has been applied suitably truncated matrix diagonalization method .
solutions depend , except the number @xmath51 of excitations ( photons ) , upon the relative size of four parameters : the resonance detuning @xmath131 , vacuum rabi - frequency @xmath132 transverse kinetic energy detuning @xmath133 and photon - induced kinetic detuning , which is the recoil energy @xmath134 in @xmath135 units @xmath136 it is convenient to introduce dimensionless parameters , scaling the energies by the recoil energy @xmath137 and all momenta - by one - photon - reemission recoil momentum @xmath138 then , as dimensionless parameter of interaction for both sets of equations appears @xmath139 the strength of coupling is essential already at @xmath140 .
let us for illustration take the sodium s atom with @xmath141 @xmath142 main transition and estimate the volume @xmath19 of microcavity for which this is so , more definitely : @xmath143 using the explicit expressions for @xmath144and @xmath137 we arrive at @xmath145 where @xmath146 and @xmath147 are the oscillator strength and the wavelength of optical transition , @xmath148 is the mass of electron . inserting the values ( @xmath149 @xmath150 @xmath151 ) we get @xmath152 note , that for cesium atom , with the analogous transition @xmath153 and for the same volume of microcavity , the coupling strength is more than one order stronger , @xmath154 figures 1a and 1b show the energy spectrum as a function of momentum for cpw and sw - cases respectively . in both cases
@xmath155 @xmath156 @xmath157 and the atom of @xmath158 is chosen .
the comparison of figures clearly indicates the mentioned differences for energy spectrums : first , the number of energy branches is finite ( @xmath159 @xmath111 @xmath160 @xmath161 ) for cpw - case , whilst it is infinite for sw - case ( are presented only some lower branches ) ; second , the cpw - spectrum is not repeated as @xmath162 -wide zones , pointing out the failure of the quasimomentum concept in this case .
the figure [ fig.2 ] , plotted for the case @xmath163 , shows , as it was expected , approaching of the cpw - picture to the sw one .
the flattening of lower curves signifies the transition of these states into the bounded - like states , for increased number of external photons .
one can see , that the gaps between the energy zones are essentially wider in cpw - case .
also is seen the well known regularity , that the widths of the gaps between the @xmath164 and ( @xmath165 bands diminish with number @xmath166 @xcite .
probability amplitude distributions are shown in figures 3a , 3b and 4a , 4b , respectively for @xmath167 and @xmath163 cases . as is seen from comparison of the graphs , in the first pair ( @xmath167 full quantum range )
there is an apparent discrepancy between cpw and sw - cases . in the second pair of graphs ( @xmath168 )
it is suppressed , but yet stays visible , even essential .
the latter graphs also display the difference of momentum distributions for bounded - like ( lower - laying levels ) and free - like ( upper laying levels ) states .
the momentum distribution in bounded - like stationary states has a two - peak form .
the peaks retire from each other for increasing energies , gradually being broader and losing in height . in free - like states ,
in contrast , the distributions are centered at the nonperturbed values ( i.e. at @xmath169 in sw - case ) . to avoid the eventual misunderstanding of depicted graphs
, it needs to be noted that the momentum axes ( abscissa ) in these figures , as well as in the following ones for probability amplitude distributions , really contains only discrete values . only for convenience of exhibition
the neighboring points in graphs have been merged via the straight lines
in this section we examine the case , where the resonance detuning @xmath170 is much larger than any other above mentioned frequency ; @xmath171 and @xmath172 this case usually is regarded as adiabatic elimination , or off - resonance approximation @xcite . in zero order for @xmath131 we obtain from ( [ 15b ] ) and ( [ 18b ] ) the equations @xmath173 @xmath174 , \eqnum{21 } \label{21}\ ] ] @xmath175 @xmath176 , \eqnum{22 } \label{22}\ ] ] @xmath99 respectively .
this approximation , widely used in problems without the kinetic energy operator in hamiltonian , calls for some comments here .
the matter is , that for instance in sw - case , the term @xmath177 already can exceed the any preliminarily given value @xmath178 .
but it has to be taken into account , that the probability amplitudes , rather out of two - peaked range of distribution , are extremely small and can be put out of consideration
. then we arrive at the condition @xmath179 for implementation of the approximation , the same as the one without the kinetic energy operator : detuning of the resonance must be much greater than the optically induced inhomogeneous width of the transition . by means of ( [ 21 ] ) and ( [ 22 ] ) the eqs .
( [ 15a ] ) and ( [ 18a ] ) are being transformed into the recurrence equations only among the ground - level atomic amplitudes : @xmath180a_{r}=\frac{\hbar \omega _ { 0}^{2}}{8\varepsilon } ( \sqrt{r(n - r+1)}a_{r-1}+na_{r}+\sqrt{(r+1)(n - r)}% a_{r+1 } ) , \eqnum{24 } \label{24}\ ] ] @xmath181a_{l}=\frac{\hbar \omega _ { 0}^{2}n}{16\varepsilon } ( a_{l-1}+2a_{l}+a_{l+1 } ) .
\eqnum{25 } \label{25}\ ] ] this approximation preserves the nature of conjunctions between the neighboring ground level amplitudes : @xmath182 . therefore , we would expect that this approximation does not cause qualitative changes for ground level stationary states .
but it , as a matter of fact , strongly depresses the excited - levels , created from the excited - level atomic states .
the probability amplitudes , calculated on the basis of eqs .
( [ 24 ] ) and ( [ 25 ] ) , are plotted in figures 5a and 5b .
all the parameters are like in figures 4a and 4b and have to fulfill the condition ( [ 23 ] ) ( @xmath183 ) : note that the chosen value @xmath184 is even larger than the natural linewidth of envisioned optical transition .
unfortunately , even approximated eq .
( [ 24 ] ) ( or ( [ 25 ] ) ) can not be solved analytically due to @xmath185 terms , quadric relative to variable @xmath102 ( or @xmath186 ) .
to proceed , we make a new supplementary approximation , regarded to as a tavis - cummings - type approximation @xcite . presuming the @xmath187-amplitude as a slowly varying function of @xmath102 @xmath188 , a permissible step for @xmath189 and if other parameters are out of bragg and doppleron resonance conditions , this approximation envisages the variable @xmath102 @xmath190 as a continuous one .
it should be expected , of course , to be a good approximation for @xmath187 , when the momentum distribution is rather dissipated .
expanding the @xmath191 in a taylor series up to the second order , @xmath192 and substituting this expression into the eq .
( [ 24 ] ) ( cpw - case ) we get a second - order ordinary differential equation for the seeking probability amplitudes and energy values @xmath193 \right\ } a(x)=0 , \eqnum{27 } \label{27}\ ] ] where @xmath194 @xmath195 and the new amplitude @xmath196 is introduced by means of a relation @xmath197 the respective equation in sw - case is @xmath198 \right\ } \text { } a(x)=0 , \eqnum{29 } \label{29}\ ] ] where @xmath199 @xmath200 being the same as in the previous case .
the classical analog of @xmath200 is the average potential energy of atomic c.m . in the light field .
note , that the resulting eqs .
( [ 27 ] ) and ( [ 29 ] ) are no longer equivalent to the preliminary eqs.([24 ] ) and ( [ 25 ] ) and some specific quantum features of the atomic states may be lost ( see in particular the last paragraph of section iv ) .
let us first consider the more simple sw - case ( eq .
( [ 29 ] ) ) .
we restrict ourselves with the so called red - detuning case when @xmath201 @xmath202 then the eq.([29 ] ) is the equation for parabolic cylinder function @xmath203 with a real argument .
its two independent solutions are @xcite @xmath204 where @xmath205 @xmath206 the requirement for @xmath207 to be limited for any values of @xmath208 ( variable @xmath186 ) leads to a condition @xmath209 .
then in accordance with ( [ 32 ] ) , the energy gets only discrete values and is determined by the formula @xmath210 simultaneously both relevant parabolic cylinder functions turn to a hermite polynomial @xmath211 and as a sequence @xmath212 , \eqnum{34 } \label{34}\ ] ] where @xmath213 is a normalizing constant .
it is worth to note that the size of quantization in ( [ 33 ] ) exactly copies the known frequency of time oscillations in the problem of atomic diffraction in the classical standing - wave field , taking into account the c.m .
motion along the field direction . in cpw - case , in distinction with sw - one
, we arrive at @xmath214 and @xmath215 } { \sqrt{r!(n - r)!}}h_{n}\left ( \frac{y(r)}{\sqrt{2}}\right )
\exp % \left [ -\frac{y^{2}(r)}{4}\right ] , \eqnum{36 } \label{36}\ ] ] where @xmath216 cpw energy spectrum ( [ 35 ] ) has two distinguishing features in respect to sw spectrum ( [ 33 ] ) .
the first is , that the total momentum @xmath69 enters into the expression of energy .
this means that the atomic c.m .
motion is not bounded in stationary states in general ( @xmath217 ) , but keeps some one - sided mean velocity .
states become bounded only in the limit @xmath218 that is , in classical field limit . as a scale of this limit
is claimed the ratio @xmath219 taking into account the definition of @xmath200 ( just after eq.([27 ] ) ) , the mentioned condition may be written in the form @xmath220 provided , of course , the condition ( [ 23 ] ) .
for @xmath221 cavity the condition ( [ 38 ] ) is satisfied at @xmath222 for sodium atoms and at @xmath223 for cesium atoms .
the second difference of ( [ 35 ] ) relative to ( [ 33 ] ) is the size of quantization , emerging by the additional term @xmath224 in square root expression .
it has negligible influence in the viewed off - resonance approximation .
evident differences occur in probability amplitudes .
it is convenient to image the expression in ( [ 36 ] ) as a product of two parts ; first , including the hermite polynomial and second , the ratio of exponential to the square root of factorials .
first part coincides with the ( [ 34 ] ) , but is shifted relative it by @xmath225 this shift has the same coefficient at @xmath226 as the kinetic energy term in ( [ 35 ] ) , and is not small in quantum optics range ( @xmath227 for off - resonance approximation ) .
the second part , i.e. the additional ratio , is a symmetric function relative to point @xmath228 , the form of which is presented in fig.[fig.6 ] ( for some values of @xmath51 ) .
as is seen , it can be viewed as unity in central region and decreasing near the edges of the definition range .
hence , the amplitude distribution in cpw - case , being product of two symmetric by itself , but shifted with respect to each other functions , is not , already , a symmetric one , as it was in cpw - case .
it is symmetric only if @xmath229 , but even in this case , the cpw distribution is somewhat narrower than the sw distribution . in analogy with energy spectrum ,
all differences disappear in the limit @xmath230 where the counterpropagating light waves appear to a far detuned two - level atom as a sinusoidal potential , just similar to the standing light wave .
figures 7a and 7b show the probability distributions calculated by means of formulas ( [ 36 ] ) and ( [ 34 ] ) respectively .
all the parameters are like in figures 5a and 5b .
comparison of fig.7a with fig.7b shows a small , almost imperceptible narrowing in each cpw-mountain range .
the mutual comparison of figures 7a and 7b with figures 5a and 5b corroborates the tavis - cummings - type approximation to be an acceptable tool in the range of bounded states .
but it totally puts out of consideration all free or quasi - free states .
in fig.8a we present a similar to fig.7a distribution , calculated for conditions more precisely satisfying tavis - cummings - type approximation ( @xmath231 ) .
its coincidence with the original , see fig.8b , is of course better .
however , it is to be memorized that the rising of @xmath51 drifts the theory to the semiclassical domain of interactions .
let us return to the theme of stationary states in the cases under consideration . our objective is to clarify the ideas about the quantum contents of interaction mechanisms in sw and cpw - cases , to indicate the differences between them and already from this point , to understand the formation of stationary states from initially free states
. to present them , we will first consider the sw - case .
the variation of atomic c.m .
motion in this case can be imagined as a diffraction of atomic matter waves on frozen
immovable object : the standing wave grating .
this is a fully classical - like phenomenon , provided by the matter - wave representation for atomic c.m .
the picture , that the scattering is a result of absorption and reemission of sw - photons should be regarded as a defective , since it does not answer to the question how the atomic momentum is changing due to these zero - momentum photons .
this obstacle is overcame in the theory by the fact that the probability amplitude distribution @xmath232 with @xmath233 , as well as the amplitude distribution @xmath234 represent only the atomic states and do not concern to photons . to realize the adiabatic evolution of the system from initially free state to the stationary coupling one
, we assume the parameter @xmath16 of the interaction to be a slow function of time , replace @xmath79 by @xmath235 in ( [ 25 ] ) ( in off - resonance approximation ) and view the @xmath232-amplitudes and the parameter @xmath79 as a slowly varying in time quantities .
after a phase transformation @xmath236 , \eqnum{40 } \label{40}\ ] ] we arrive at the equation @xmath237 for density matrix elements , which are more preferable for the putted objective . here
@xmath238 is the analog of doppler frequency shift .
the equation ( [ 41 ] ) is , of course , too complicated to be inspected in general form , but we may consider the initial stage of interaction , taking in addition @xmath239 then @xmath240 are fully imaginary with @xmath241 @xmath242 ( for @xmath243 ) and the relative rate of depletion of the state with @xmath244 and @xmath245 is given by relative values of @xmath246 and @xmath247 respectively . if @xmath248 then @xmath249 and less modulo momentum states are being filled more rapidly than more modulo momentum states . as a result ,
the mean atomic momentum decreases gradually , gathering for asymptotically long times about the zero value . in attained stationary state , the mean atomic momentum approaches zero . note , that this is a stimulated analog of well known doppler cooling ( slowing ) of red - detuned atoms in the standing wave .
the case of cpw we will start ( and mainly restrict ourselves ) by the immovable atom model ( about commutative set of operators see in appendix ) , since it gives an opportunity to present the additional mechanism for stationary state formation .
putting @xmath250 in eqs .
( [ 15a ] ) and ( [ 15b ] ) and substituting the second into the first we readily arrive at equation @xmath251}\left [ \sqrt{r(n - r+1)}% % tcimacro{\func{im}}% % beginexpansion \mathop{\rm im}% % endexpansion ( \alpha _ { r}\alpha _ { r-1}^{\ast } ) + \sqrt{(r+1)(n - r)}% % tcimacro{\func{im}}% % beginexpansion \mathop{\rm im}% % endexpansion ( \alpha _ { r}\alpha _ { r+1}^{\ast } ) \right ] \eqnum{42 } \label{42}\ ] ] for density matrix elements , where @xmath252}\right ) t% \right ] .
\eqnum{43 } \label{43}\ ] ] here , in contrast to sw - case , each atomic c.m .
state is single - bonded with the pair of photon numbers in cpw .
the mathematical response of this connection is that the @xmath253and @xmath254 amplitudes are the representation of both atomic c.m . and photonic states simultaneously .
the difference between the rates of @xmath255 and @xmath256 transitions now is determined by square roots @xmath257 and @xmath258 the former term represents the process of reemission of a photon from the first travelling wave , containing preliminary @xmath102 photons , into the second one , with respectively @xmath259 photons , concomitantly altering the atomic momentum in @xmath260 the second term represents the inverse process , i.e. the reemission of a photon from the second wave into the first wave , accompanying it with a @xmath162 atomic momentum change .
the physical reason of differences between the mentioned rates is solely the bosonic nature of photons .
direct comparison shows the dominance of reemission of photons from more photon
wave into the less photon wave , which finally results in equalization of mean photon numbers in both counterpropagating travelling waves @xcite .
hence , in attained stationary states the mean field momentum approaches zero .
thus , both analyzed cases possess diverse mechanisms of approaching to stationary state , but the final results are similar in the sense , that since one of the coupling subsystems , atom or field , is immovable or frozen , the mean momentum of the other subsystem unavoidably approaches zero . speaking in fancy form ,
the immovable part of the system takes on itself the surplus of the momentum from the other subsystem . a clear and expected result . in cpw - case with a moving atom , however , this result is not the general case .
the reason is the conservation of system s momentum . in this case
the system evolves under action of both mechanisms .
first of them , the diffraction , tends to suppress the mean atomic momentum , whilst second one , the bosonic nature , in contrary , tends to suppress the difference between the mean numbers of counterpropagating travelling waves .
since they are not possible simultaneously ( except the case @xmath229 ) , in the course of time they mature the stationary states with definite , asymmetric in general , statistical distributions , where these tendencies are mutually compatible .
the role of bosonic mechanism however is essential in the quantum - optic domain ( @xmath261 ) and gradually diminishes out of it . to conclude this section
, we would like to make a notice about one unavoidable peculiarity of tavis - cummings - type approximation , with respect to adiabaticity of interaction . taking the photon momentum as a continuous variable ,
we have suppressed the size of its quantum up to zero ( as small as possible ) .
thereby we make feasible the arbitrary small variations in the energy of atomic c.m . motion . as a consequence ,
any as small as possible variation of the field intensity ( more concrete of parameter @xmath16 ) should cause changes in atomic c.m .
motion and breaks down the adiabatic approximation . because of this , the approximate solutions ( [ 34 ] ) and ( [ 36 ] ) , in contrast to exact eqs .
( [ 15a ] ) , ( [ 15b ] ) or ( [ 18a ] ) , ( [ 18b ] ) , can not be adiabatically related to the free system s state .
continuous quantum transition in atomic c.m .
motion completely washes out the information about the initial state and its energy .
really , the switching off of interaction ( @xmath262 ) in tavis cummings - type solutions leads to only one @xmath263 value of energy and to one corresponding probability amplitude with @xmath264 photons in the first travelling wave and the rest @xmath265 photons in the second travelling wave .
it is necessary , however , to clearly realize that the parameters @xmath76 and @xmath38 preserve their physical essence as parameters of initial state ; they are determined by the relations ( [ 7 ] ) , ( [ 10 ] ) and ( [ 13 ] ) ( in cpw - case ) and are @xmath16-independent .
simply , the tavis - cummings - type approximation interrupts any , even adiabatic , connection of the stationary state with the initial one .
the parameters @xmath76 and @xmath38 can be used in this approximation as externally ( additionally ) given ones , and in case of necessity can be corroborated by means of exact equations .
these resonances , bragg and doppleron , have been identified for an atom moving in a standing wave @xcite and have got implementations in atom optics and laser cooling @xcite . here
we would consider only the limit of small intensities , focusing on the role of multiphoton recoil energy .
some preliminary words , for reminding , how these resonances are being extracted from equations .
thinking of the limit @xmath262 as a leading one to the initial state , one should put a corresponding condition on the initial state amplitudes .
for example , if @xmath266and @xmath59 photons in cpw and the atom is on the ground level , then @xmath267 as a consequence , we get from ( [ 15a ] ) the value of the free system energy , @xmath268 here the atomic momentum @xmath38 is arbitrary .
when in contrary , the limit @xmath269 is thought of as a transition to the final free state , we may only confirm the equality of energy to its ( [ 44 ] ) free value .
it follows from the conservation of energy . also taking into account the relations ( [ 7 ] ) and ( [ 10 ] ) , from eqs .
( [ 15a ] ) , ( [ 15b ] ) we directly arrive at @xmath270 or @xmath271 the solution with @xmath272 and respectively @xmath273 returns the system into the initial state .
the next solution with @xmath274 gives a sole nonzero value for @xmath275 ground - level amplitude .
corresponding to this case final atomic momentum ( see the wave function ( [ 11 ] ) ) @xmath276 which is the condition for bragg scattering .
the third possible solution of ( [ 45 ] ) and ( [ 46 ] ) occurs at @xmath277 and has a sole nonzero solution for the corresponding excited - level amplitude @xmath56 .
this is the case of doppleron - resonance scattering , when the atom leaves the zone of interaction on excited energy level @xcite .
let us stop our attention on doppleron - resonance condition ( [ 48 ] ) in order to examine the behavior of final momentum @xmath278 versus the initial momentum @xmath38 . to this
end the condition ( [ 48 ] ) may be viewed as a dependance of @xmath102 on @xmath38 and then put into the excited - atom final momentum expression @xmath279 computation gives @xmath280 where the atomic initial momentum satisfies the doppleron resonance condition ( [ 48 ] ) . for
a fixed frequency value @xmath4 the allowed values of initial momentum @xmath38 constitute a discrete set , corresponding to the integer values of @xmath281 in ( [ 48 ] ) .
graphs of ( [ 49 ] ) are plotted in fig.[fig.9 ] . for comparison we also plot the graphs without the recoil energy term ( the last one in ( [ 48 ] ) ) .
as we can see , the presence of recoil energy dramatically changes the character of @xmath282 dependance .
the first distinctive feature is in splitting of possible values of @xmath278 for every @xmath38 into two branches , herewith converging the high - order resonances ( @xmath283 ) @xmath284to the bragg resonances ( @xmath285 in the additional branch .
the second difference is the lack of resonances in small atomic momentum range ( @xmath286 ) and connected with it , the convergence of @xmath278 to zero at the limiting point with @xmath287 it means that the inherent possibilities of the doppleron - cooling really are more than it follows from the theory , not including the recoil energy term .
we analyze the stationary states of jcm with sw and cpw fields , treating the external , as well as the internal , atomic coordinates quantum - mechanically .
the inspection shows essential deviations between sw- and cpw - cases in quantum optics range ( @xmath288 ) , which first of all have qualitative character . as a such matter
can be noted the deterioration of the concept of quasimomentum , the asymmetry and narrowing of momentum distribution , the coincidence of photonic and atomic momentum distributions in cpw - case .
since the physical reason of the mentioned differences is the existence of the momentum conservation , it can be concluded that the modern cavity qed an the theory of measurement , being developed on the base of quantized sw - cavities , in fact have only dealt with the conservation of energy .
the cavity qed and related problems for ring cavities , where the momentum conservation exists too and has to alternate the state - pattern in
atom+field
system , need a performance , yet .
moreover , the experimental confirmation just of cpw - case regularities , stationary or not , may be adopted as the direct nonmediated evidence for the photon momentum to be quantized .
reexamination of kinematic resonances shows that due to recoil energy term the doppleron - resonance splits into two branches and , which is more important , can not take place in the range of small incident kinetic energies ( where @xmath289 ) .
this work was supported by istc grant a - 215 - 99 .
the case of immovable atom ( @xmath290 ) is a separate one for stationary state problem and its analysis is not logical on the basis of eqs .
( [ 15a ] ) , ( [ 15b])@xmath136 the reason for such a separation is that the @xmath62 and @xmath29 operators loose the inherent physical contents and must be omitted . instead of them
enters a new , commutative with @xmath291 and @xmath292 photonic operator @xmath284@xmath293 which has the eigenstates @xmath294 @xmath295 it is referred to as an operator of interference , in the sense that its eigenvalues play the same role in the quantized - field picture as the term @xmath296 in the classical field picture of interaction @xcite .
s. haroche and j.m .
raimond , adv .
20 , 347 ( 1985 ) ; h. yoo and j.h .
eberly , phys .
118 , 239 ( 1985);p .
meystre , phys . rep .
219 , 243 ( 1992 ) ; a.i .
lvovsky , h. hansen , t. aichele , o. benson , j. mlynek and s. schiller , phys.rev.lett .
87 , 050402 ( 2001 ) .
cook and a.f .
bernhardt , phys .
a 18 , 2533 ( 1978 ) ; a.p .
kazantsev and g.i .
surdutovich , sov . phys .-
pisma jetp 31 , 542 ( 1980 ) ; p. meystre , e. schumacher and s. stenholm , opt .
73 , 443 ( 1989 ) ; a.zh .
muradyan , sov phys .- izvestia an arm .
ssr , fizika 10 , 361 ( 1975 ) ; p.j .
martin , p.l .
gould , b.g .
oldaker , a. miklich and d.e .
pritchard , phys .
rev . a 36 , 2495 ( 1987 ) ; g.s .
agarwal and k. moelmer , phys .
a 47 , 5158 ( 1993 ) ; m. k. oberthaler , r. abfalterer , s. bernet , c. keller , j. schmiedmayer and a. zeilinger , phys . rev . a 60 , 456 ( 1999 ) ;
b. ovchinnikov , j.h .
muller , m.r .
doery , e.j.d .
vredenbregt , k. holmerson , s.l . rolston and w.d .
phillips , phys.rev .
83 , 284 ( 1999 ) .
jaynes and f.w .
cummings , proc ieee 51 , 89 ( 1963 ) ; j.h .
eberly , n.b .
narozhny and j.j .
sanchez - mondragon , phys .
33 , 1323 ( 1980 ) ; g. rempe , h. walter and n. klein , phys . rev . lett .
58 , 353 ( 1987 ) ; v. buzek , h. moya - cessa and p.l .
knight , phys .
a 45 , 8190 ( 1992 ) ; b.w .
shore and p.l .
knight , j. mod .
40 , 1195 ( 1993 ) ; s.m .
barnet and p.m. radmore , methods in theoretical quantum optics ( clarendon press , oxford , 1997 ) .
g. compagno , j.s .
peng and f. persico , phys . rev a 26 , 2065 ( 1982 ) m. lewenstein , l. you , j. cooper and k. burnett , psys.rev .
a 50 , 2207 ( 1994 ) ; m. freyberber , a.m. herkommer , d.s .
krahmer , e. mayr and w.p .
schleich , adv.atom .
41 , 143 ( 1999 ) .
p. marte , p. zoller and j.h .
hall , phys.rev .
a 44 , r4118 ( 1991 ) ; j. lawall and m. prentiss , phys .
72 , 993 ( 1994 ) , l.s .
goldener , c. gertz , r.j.c .
spreeuw , s.l .
rolston , c.i .
westbrook and w.d .
phillips , phys .
72 , 994 ( 1994 ) ; p.d .
featonby , g.s .
summy , j.l .
martin , h.wu , k.p .
zetie , c.j .
foot and k. burnett .
a 53 , 373 ( 1996 ) . | the eigenstate problem of the jaynes - cummings model on the basis of complete hamiltonian , including the center - of -mass kinetic energy operator , is treated .
the energy spectrum and wave functions in standing - wave ( sw)- and counterpropagating waves ( cpw)- cases are calculated and compared with each other .
it is shown that in cpw - case i ) the atomic momentum distribution is asymmetric and somewhat narrower in general ; ii ) the concept of quasimomentum is not applicable and instead the ordinary momentum concerns the problem ; iii ) atomic and photonic state distributions are self - consistent , and , in consequence iiii ) mean number of photons in the counterpropagating traveling waves and mean atomic momentum match .
explicit analytic expressions for energy eigenvalues and eigenfunctions are found in tavis -cummings - type approximation [ phys .
rev .
170 , 379(1968 ) ] and is pointed , that it implies only the bounded - like states for atomic center - of - mass motion .
it is also shown that if the recoil energy is taken into account , the doppleron resonance is split into two branches , one of which diverges to bragg - like resonance in the high - order range . |
in the standard big bang model of nucleosynthesis ( sbbn ) , the light isotopes d , @xmath7he , @xmath3he and @xmath8li were produced by nuclear reactions a few minutes after the birth of the universe .
given the number of light neutrino species @xmath9 = 3 and the neutron lifetime , the abundances of these light elements depend on one cosmological parameter only , the baryon - to - photon ratio @xmath10 , which in turn is directly related to the density of ordinary baryonic matter @xmath11 .
the ideal objects for determination of the primordial helium abundance are blue compact dwarf ( bcd ) galaxies .
these dwarf systems are the least chemically evolved galaxies known , so they contain very little helium manufactured by stars after the big bang .
because the big bang production of @xmath3he is relatively insensitive to the density of matter , the primordial abundance of @xmath3he must be determined to very high precision ( better than a few percent relative accuracy ) in order to put useful constraints on @xmath12 .
this precision can be achieved with very high signal - to - noise ratio optical spectra of bcds .
these bcds are undergoing intense bursts of star formation , giving birth to high excitation supergiant h ii regions , which allow an accurate determination of the helium abundance in the ionized gas through the bcd s emission - line spectrum .
the primordial helium mass fraction @xmath13 of @xmath3he is usually derived by extrapolating the @xmath4 o / h and @xmath4 n / h correlations to o / h = n / h = 0 , as proposed originally by peimbert & torres - peimbert ( 1974 , 1976 ) and pagel , terlevich & melnick ( 1986 )
. many attempts at determining @xmath13 have been made , using these correlations on various samples of dwarf irregulars and bcds ( e.g. , pagel et al .
1992 ; izotov , thuan & lipovetsky 1994 , 1997 , hereafter itl94 and itl97 ; olive , steigman & skillman 1997 ; izotov & thuan 1998 ; pagel 2000 ) .
another way to infer the primordial @xmath3he abundance is to measure the helium abundance in the most metal - deficient bcds , which is very close to the primordial value ( e.g. , izotov et al .
1999 ) .
recent @xmath13 determinations made by izotov & thuan ( 1998 ) and izotov et al .
( 1999 ) resulted in a very narrow range of @xmath13 @xmath5 0.244 0.245 .
they used high signal - to - noise spectroscopic observations of bcds reduced in a homogeneous way .
a self - consistent method was applied to correct he i emission line intensities for the collisional and fluorescent enhancement mechanisms which lead to deviations from the values predicted by recombination theory .
the use of several he i lines allows one to discriminate between collisional and fluorescent enhancements which change the line intensities in different ways .
it also allows one to estimate the importance of underlying stellar absorption in each galaxy and to improve the precision of the @xmath13 determination through the use of several lines .
the details of this approach are discussed by itl97 and izotov & thuan ( 1998 ) .
our present study is the continuation of helium abundance determinations in the lowest - metallicity bcds based on spectroscopic observations with the 10 m keck telescope .
earlier such studies have been done for the bcds sbs 0335052 ( @xmath1/40 , izotov et al .
1999 ) and sbs 0940 + 544 ( @xmath1/27 , guseva et al .
we focus here on two southern bcds , tol 1214277 and tol 65 @xmath14 tol 1223359 , the very low metallicity of which ( @xmath5 @xmath1/25 ) has been established by earlier spectroscopic work ( kunth & sargent 1983 ; campbell et al .
1986 ; pagel et al . 1992 ; masegosa , moles & campos - aguilar 1994 ; fricke et al .
the motivation of our work is as follows .
first , these galaxies are relatively bright targets for a large telescope which allows the derivation of the @xmath3he abundance with great precision .
second , tol 1214277 and tol 65 are the second and third lowest - metallicity bcds in the pagel et al .
( 1992 ) sample ( after i zw 18 ) , for which those authors derived very low @xmath3he mass fractions @xmath4 = 0.233 and 0.231 respectively .
those values are in disagreement with the significantly larger @xmath4 s derived in later studies of other very low metallicity bcds .
third , recently several papers have appeared ( e.g. , ballantyne , ferland & martin 2000 ; viegas , gruenwald & viegas 2000 ; sauer & jedamzik 2001 ; peimbert , peimbert & luridiana 2001 ; stasiska & izotov 2001 ) where systematic effects on the @xmath3he abundance determination are discussed .
here we estimate the systematic uncertainties for several of the best - observed low - metallicity bcds , including tol 1214277 and tol 65 . in sect .
2 we describe the observations and data reduction .
heavy element abundances are derived in sect .
the results of the @xmath3he abundance determination are discussed in sect .
we discuss the systematic uncertainties in @xmath3he abundance determinations in sect .
the summary is in sect .
the keck ii spectroscopic observations of tol 1214277 and tol 65 were carried out on january 9 , 2000 , with the low - resolution imaging spectrograph ( lris ) ( oke et al .
1995 ) , using the 300 groove mm@xmath15 grating , which provides a dispersion 2.52 pixel@xmath15 and a spectral resolution of about 8 in first order .
the slit was 1@xmath16180 , centered on the brightest central regions and oriented with a position angle p.a .
= 19@xmath17 for tol 1214277 and 51@xmath17 for tol 65 ( fig .
[ fig1 ] ) .
no binning along the spatial axis has been done , yielding a spatial sampling of 02 pixel@xmath15 .
the total exposure time was 45 min for each galaxy , broken into three 15-min exposures .
all exposures were taken at airmasses of 1.7 and 1.8 for tol 1214277 and tol 65 , respectively .
the seeing was 09 .
no blocking filter was used .
therefore some second - order contamination is present in the red parts of spectra at wavelengths @xmath18 7000 .
the spectrophotometric standard stars feige 34 and hz 44 were observed for flux calibration .
spectra of a hg - ne - ar comparison lamp were obtained before and after each observation to provide the wavelength calibration .
data reduction of the observations was carried out at the main astronomical observatory of ukraine using the iraf software package .
this reduction included bias subtraction , cosmic - ray removal and flat - field correction using exposures of a quartz incandescent lamp .
after wavelength calibration , night - sky background subtraction , and correction for atmospheric extinction , each frame was calibrated to absolute fluxes . because both bcds were observed at large airmasses , the atmospheric differential refraction can be important ( filippenko 1982 ) .
this effect is smaller for tol 1214277 because of its higher declination and slit position angle close to the parallactic one . to minimize the effect of the atmospheric differential refraction we extracted one - dimensional spectra from large apertures 1 @xmath16 68 for tol 1214277 and 1 @xmath16 84 for tol 65 .
the spectra are shown in figs .
[ fig2 ] and [ fig3 ] .
they are dominated by very strong emission lines .
remarkable spectral features in tol 1214277 are the strong nebular he ii @xmath24686 and [ fe v ] @xmath24227 emission lines suggesting a very hard stellar radiation field in the bcd ( fig . [ fig2 ] ) .
the latter line was detected first by fricke et al .
( 2001 ) in the vlt spectrum of tol 1214277 .
this line is absent in the spectrum of tol 65 where he ii @xmath24686 is also weaker , implying that the radiation in this galaxy is softer .
the fluxes of the nebular lines have been measured by fitting gaussians to the line profiles .
the errors in the line fluxes include the errors in placement of the continuum and those in the gaussian fitting .
we also take into account the errors introduced by uncertainties in the spectral energy distributions of the standard stars .
standard star flux deviations for both feige 34 and hz 44 are taken to be 1% ( oke 1990 ; bohlin 1996 ) .
these 1@xmath6 errors have been propagated in calculations of the electron temperature , electron number density and elemental abundances .
the observed and extinction - corrected emission line fluxes and observed equivalent widths are shown in table [ tab1 ] together with the extinction coefficient @xmath19(h@xmath20 ) , derived from the decrement of hydrogen emission lines which includes both milky way and target galaxy extinction , the absolute flux @xmath21(h@xmath20 ) of the h@xmath20 emission line and average equivalent width @xmath22(abs ) of the hydrogen absorption lines .
the errors of the fluxes and equivalent widths of the emission lines introduced by second - order contamination are relatively small .
we estimate this effect by measuring the fluxes of the [ o ii ] @xmath23727 and [ ne iii ] @xmath23868 emission lines in the second order spectra of both galaxies and find that they are respectively @xmath5 1.5% and @xmath5 3 4% of those in the first order spectra .
this implies that the effect of the second - order contamination is less than 1% at wavelengths shorter 7500 , smaller than the flux errors of the weak lines seen at 7000 7300 .
therefore , we do not take into account this effect .
to derive heavy element abundances , we have followed the procedure detailed in itl94 and itl97 .
we adopted a two - zone photoionized h ii region model ( stasiska 1990 ) including a high - ionization zone with temperature @xmath23(o iii ) , and a low - ionization zone with temperature @xmath23(o ii ) .
we have determined @xmath23(o iii ) from the [ o iii]@xmath24363/(@xmath24959+@xmath25007 ) ratio using a five - level atom model . that temperature is used for the derivation of the o@xmath24 , ne@xmath24 and ar@xmath25 ionic abundances . to derive @xmath23(o ii ) , we have utilized the relation between @xmath23(o ii ) and @xmath23(o iii ) ( itl94 ) , based on a fit to the photoionization models of stasiska ( 1990 ) . the temperature @xmath23(o ii ) is used to derive the o@xmath26 , n@xmath26 , s@xmath26 and fe@xmath26 ionic abundances . for ar@xmath24 and s@xmath24
we have adopted an electron temperature intermediate between @xmath23(o iii ) and @xmath23(o ii ) following the prescriptions of garnett ( 1992 ) . the electron number density @xmath27(s ii ) ( table [ tab2 ] )
is derived from the [ s ii ] @xmath26717/@xmath26731 ratio .
we point out that the flux of the [ s ii ] @xmath26717 emission line in tol 1214277 spectrum is significantly reduced due to coincidence with a night sky absorption line .
this artificially low intensity results in a significant overestimate of @xmath27(s ii ) . however , the heavy element abundances in low - density h ii regions do not depend on @xmath27 .
therefore , the uncertainties in @xmath27(s ii ) ( table [ tab2 ] ) do not contribute significantly to the error budget of the heavy element abundances .
the oxygen abundance is derived as @xmath28 where the o@xmath25 abundance is derived from the relation @xmath29 total abundances of other heavy elements were computed after correction for unseen stages of ionization as described in itl94 and thuan , izotov & lipovetsky ( 1995 ) .
the abundances of oxygen and other heavy elements obtained in this study for both galaxies are in general agreement with previous studies .
our value for the oxygen abundance in tol 1214277 , 12 + log ( o / h ) = 7.54 @xmath0 0.01 , compares with values of [email protected] ( campbell et al .
1986 ) , [email protected] ( pagel et al .
1992 ) , [email protected] ( masegosa et al .
1994 ) , 7.58@xmath30 ( kobulnicky & skillman 1996 ) , [email protected] ( fricke et al .
in tol 65 we derive an oxygen abundance 12 + log ( o / h ) = 7.54 @xmath0 0.01 which is in agreement with [email protected] ( kunth & sargent 1983 ) , [email protected] ( campbell et al .
1986 ) , [email protected] ( pagel et al . 1992 ) , 7.40 7.54 ( masegosa et al . 1994 ) , 7.56@xmath31 ( kobulnicky & skillman 1996 ) .
the spectral resolution of the tol 1214277 spectrum is not enough to measure the flux of the weak [ n ii ] @xmath26583 which is contaminated by the h@xmath32 emission line .
therefore , to derive the nitrogen abundance we use the [ n ii]/h@xmath32 flux ratio obtained by pagel et al .
( 1992 ) from a higher resolution spectrum .
we obtain log n / o = [email protected] .
this value is lower than previous values of [email protected] ( pagel et al .
1992 ) and 1.45@xmath33 ( kobulnicky & skillman 1996 ) . the difference in n / o comes mainly from the differing fluxes of [ o ii ] @xmath23727 emission line .
this line is used for determination of the ionization correction factor for nitrogen and it varies from @xmath34(@xmath23727)/@xmath34(h@xmath20 ) = 0.23 ( pagel et al .
1992 ) to 0.34 in this paper , the difference being 0.17 dex .
fricke et al . ( 2001 ) derived log n / o = [email protected] from vlt observations .
however , no flux calibration was available for those observations and hence fricke et al .
( 2001 ) used earlier spectroscopic observations of tol 1214277 with the kpno 2.1 m telescope to calibrate their vlt spectrum .
the [ n ii ] @xmath26583 emission line is stronger in the spectrum of tol 65 and we derive log n / o = 1.64 @xmath0 0.02 .
all other log n / o values for tol 65 are derived from a single observation by kunth & sargent ( 1983 ) who obtained log
n / o = 1.75 @xmath0 0.07 , while pagel et al .
( 1992 ) quote [email protected] , and kobulnicky & skillman ( 1996 ) quote 1.79@xmath35 .
our value of log n / o in tol 65 obtained from the high signal - to - noise ratio spectrum is the same as that derived for tol 1214277 and is very close to the mean value of 1.60 , derived for the most metal - deficient galaxies with 12 + log o / h @xmath36 7.6 ( thuan et al .
1995 ; izotov & thuan 1999 ) , further supporting the very low dispersion of the n / o ratio in those galaxies .
such a constant n / o abundance ratio in the lowest - metallicity bcds favors primary production of nitrogen in massive stars ( izotov & thuan 1999 ) and may have important implications for analysis of the abundance patterns in high - redshift damped ly@xmath32 systems ( izotov , schaerer & charbonnel 2001 ) .
the element - to - oxygen abundance ratios in tol 1214277 and tol 65 for @xmath32-product elements are the same within the errors ( table [ tab2 ] ) and they are close to the mean values derived for low - metallicity bcds ( izotov & thuan 1999 ) . the exception is iron . while the fe / o abundance ratio in tol 65 , though slightly lower , is consistent with the mean value for bcds ( izotov & thuan 1999 ) , the fe / o abundance ratio in tol 1214277 is 2@xmath6 larger .
a similar fe / o abundance ratio was found by izotov et al .
( 1997 , 1999 ) for the bcd sbs 0335052 and was interpreted to be a result of the contamination of the [ fe iii ] @xmath24658 emission line by stellar or nebular c iv @xmath24658 line emission .
this interpretation seems to be likely , because both sbs 0335052 and tol 1214277 are galaxies with the strongest nebular he ii @xmath24686 emission lines , implying very hard ionizing radiation . only in those two bcds
was the [ fe v ] @xmath24227 emission line detected , again supporting the presence of hard radiation ( fricke et al .
our derived abundance of o@xmath25 in tol 1214277 is 5.5% of the total oxygen abundance ( table [ tab2 ] ) .
hence we expect that a significant amount of carbon in the h ii region is present in the form of c@xmath37 , implying the presence of c iv @xmath24658 line emission .
he i emission - line fluxes are converted to singly ionized helium abundances @xmath38 @xmath14 he@xmath26/h@xmath26 using theoretical he i recombination line emissivities by smits ( 1996 ) .
however , collisional and fluorescent enhancements can cause the observed he i fluxes to deviate from recombination values . in order to correct for these effects ,
we have adopted the following procedure , discussed in more detail in itl94 and itl97 .
we evaluate the electron number density @xmath27(he ii ) and the optical depth @xmath39(@xmath23889 ) in the he i @xmath23889 line in a self - consistent way , so that the he i @xmath2@xmath23889/5876 , 4471/5876 , 6678/5876 and 7065/5876 line ratios have their recombination values , after correction for collisional and fluorescent enhancement .
corrections are determined using the formulae by kingdon & ferland ( 1995 ) for collisional enhancement and the izotov & thuan ( 1998 ) fits to robbins ( 1968 ) calculations for fluorescent enhancement .
the he i @xmath23889 and 7065 lines play an important role because they are particularly sensitive to both optical depth and electron number density .
since the he i @xmath23889 line is blended with the h8 @xmath23889 line , we have subtracted the latter , assuming its intensity to be equal to 0.106 @xmath34(h@xmath20 ) ( aller 1984 ) , after correction for interstellar extinction and underlying stellar absorption in hydrogen lines .
the singly ionized helium abundance @xmath38 and @xmath3he mass fraction @xmath4 is obtained for each of the three he i lines @xmath2@xmath2 4471 , 5876 and 6678 .
we then derive the weighted mean @xmath38 of these three determinations , the weight of each line being determined by its intensity .
however , this weighted mean value may be underestimated due to the lower value of @xmath38(4471 ) resulting from underlying stellar absorption .
therefore , in subsequent discussions we also use the weighted mean values of @xmath4 derived from the intensities of only two lines , he i @xmath25876 and @xmath26678 .
additionally , we have added to @xmath38 the abundance of doubly ionized helium @xmath40 which is derived from the he ii @xmath24686 emission line flux .
finally the helium mass fraction is calculated as @xmath41}{1 + 4y } , \label{eq : y}\ ] ] where @xmath42 = @xmath38 + @xmath40 is the number density of helium relative to hydrogen ( pagel et al .
1992 ) .
the results of the @xmath3he abundance determination are presented in table [ tab3 ] , where we show the adopted electron temperature @xmath23 and the derived electron number density @xmath27 in the he@xmath26 zone , the optical depth in the he i @xmath23889 emission line , ionic abundances @xmath38 and @xmath40 , total abundances @xmath42 and helium mass fractions @xmath4 derived for each line and the two weighted means .
the errors in @xmath23(o iii ) and @xmath27 ( he ii ) are propagated in calculations of the helium mass fractions @xmath4 .
the helium mass fraction in tol 1214277 is slightly lower for the he i @xmath24471 emission line , which is most subject to underlying stellar absorption .
similarly , systematically lower @xmath4 from the he i @xmath24471 emission line was derived earlier in two bcds , sbs 0335052 ( @xmath1/40 , izotov et al .
1999 ) and sbs 0940 + 544 ( @xmath1/27 , guseva et al .
2001 ) , observed with keck .
the contribution of the doubly ionized he in tol 1214277 is significant and amounts to 6% of the total he abundance .
earlier , fricke et al . ( 2001 ) arrived at the same conclusion for this bcd .
the helium mass fraction in tol 1214277 is @xmath4 = 0.2458 @xmath0 0.0039 when all three lines are used , and 0.2466 @xmath0 0.0043 when the he i @xmath24471 emission line is excluded .
the effect of the underlying stellar absorption is more significant in tol 65 .
the helium mass fraction @xmath4 derived from the he i @xmath24471 emission line is @xmath5 10% lower than that derived from other lines .
the contribution of doubly ionized helium in tol 65 is smaller than that in tol 1214277 and amounts to @xmath5 1.4% of the total helium abundance .
the helium mass fraction in tol 65 is @xmath4 = 0.2410 @xmath0 0.0050 when all three lines are used , and 0.2463 @xmath0 0.0057 when the he i @xmath24471 emission line is excluded . because of the effect of underlying absorption in the he i @xmath24471 emission line , we finally adopt for tol 1214277 and tol 65 respectively @xmath4 = 0.2466 @xmath0 0.0043 and 0.2463 @xmath0 0.0057 .
these values are very similar to 0.2463 @xmath0 0.0015 in sbs 0335052 ( izotov et al . 1999 ) and 0.2468 @xmath0 0.0034 in sbs 0940 + 544 ( guseva et al . 2001 ) derived from the analysis of high signal - to - noise ratio keck spectra .
the similarity of @xmath4 in the lowest - metallicity bcds suggests that statistical errors in the helium abundance determination are small .
however , some systematic effects may change the value of @xmath4 .
we already pointed out one such effect , the underlying stellar absorption , which if not accounted for , results in the underestimation of @xmath4 .
this effect is more pronounced for the he i @xmath24471 emission line .
recently , gonzlez delgado , leitherer & heckman ( 1999 ) have produced synthetic spectra of h balmer and he i absorption lines in starburst and poststarburst galaxies .
they predict that the equivalent width of the he i @xmath24471 absorption line for young starbursts with an age @xmath43 @xmath44 5 myr , which is the case for tol 1214277 and tol 65 , can be in the range 0.4 0.6 . comparing with equivalent widths of the he i @xmath24471 emission line 7.6 and 4.8 in tol 1214277 and tol 65 ,
we conclude that the underestimate of the he abundance can be as high as @xmath5 10% in the case of the he i @xmath24471 emission line . for the other two he i lines @xmath25876 and @xmath26678 lines
the underestimate seems to be significantly smaller because of larger emission line equivalent widths .
unfortunately , gonzlez delgado et al .
( 1999 ) did not calculate equivalent widths for the he i @xmath25876 and @xmath26678 absorption lines .
the weighted mean helium mass fraction in our calculations is mainly defined by the strongest he i @xmath25876 emission line with the highest weight .
the upward correction of @xmath4 for this line due to the underlying stellar absorption is not larger than @xmath5 1% for tol 1214277 and @xmath5 1%
2% for tol 65 .
here we assume that the equivalent width of the he i @xmath25876 absorption line is 0.4 , which is @xmath5 100 times smaller than the equivalent width of the emission line in tol 1214277 and @xmath5 80 times smaller in tol 65 ( table [ tab1 ] ) .
another source of systematic uncertainties comes from the assumption that the h@xmath26 and he@xmath26 zones in the h ii region are coincident .
however , depending on the hardness of the ionizing radiation , the radius of the he@xmath26 zone can be smaller than the radius of the h@xmath26 zone in the case of soft ionizing radiation and larger in the case of hard radiation . in the former case ,
a correction for unseen neutral helium should be made , resulting in an ionization correction factor @xmath45(he ) @xmath18 1 and hence a higher helium abundance . in the latter case ,
the situation is opposite and @xmath45(he ) @xmath36 1 .
furthermore , the electron temperature @xmath23 in the o@xmath24 zone derived from the collisionally excited [ o iii ] lines was assumed in our calculations to be constant and is the same as that in the h@xmath46 and he@xmath46 zones .
however , @xmath23(o iii ) tends to be larger than the temperature for the recombination lines of h i and he ii and , if applied , results in an overestimate of the helium abundance .
both these effects have been discussed in several studies ( e.g. , pagel et al .
1992 ; itl97 ; steigman , viegas & gruenwald 1997 ; olive et al .
1997 ; viegas et al .
2000 ; peimbert , peimbert & ruiz 2000 ; ballantyne et al .
2000 ; sauer & jedamzik 2001 ) .
it was shown that the correction of the helium abundance can be as high as several percent in either downward or upward directions depending on the hardness of the radiation .
the hardness is characterized by the `` radiation softness parameter '' @xmath10 defined by vlchez & pagel ( 1988 ) as @xmath47 besides @xmath10 some other parameters have been used to derive @xmath45(he ) , in particular the [ o iii ] @xmath25007/h@xmath20 and [ o iii ] @xmath25007/[o i ] @xmath26300 emission line flux ratios ( ballantyne et al . 2000 ) , the ionization parameter @xmath48 , which is the ratio of ionizing photon density to gas density , and the combination of all preceeding parameters ( sauer & jedamzik 2001 ) . sauer & jedamzik ( 2001 ) calculated an extensive grid of photoionized h ii region models aiming to derive the correction factors as functions of @xmath10 and @xmath48 .
their conclusion was that a downward correction of @xmath4 as much as 6% and 2% is required respectively for ionization parameters log @xmath48 = 3.0 and 2.5 ( see their fig .
18 ) . however , the downward correction is @xmath44 1% if log @xmath48 @xmath49 2.0 . for tol 1214277 and tol 65
we find respectively log @xmath10 = 0.29 and 0.08 .
the ratios ( [ o iii ] @xmath24959 + 5007)/[o ii ] @xmath23727 of 19.9 and 7.2 in those galaxies ( table [ tab1 ] ) at an oxygen abundance 12 + log o / h = 7.54 correspond to an ionization parameter log @xmath48 @xmath49 2.0 ( mcgaugh 1991 ) . in particular , campbell ( 1988 ) derived log @xmath48 = 1.61@xmath50 for tol 1214277 at the 70% confidence level . with these @xmath10 and @xmath48 values ,
the downward correction of @xmath4 due to ionization effects and variations of the electron temperature in tol 1214277 and tol 65 is unlikely to be greater than @xmath5 1% .
taking into account the fact that the upward correction of @xmath4 due to underlying stellar absorption can be as high as 1% 2% , we conclude that both effects seem to offset each other and the combined systematic uncertainty is @xmath44 1% in tol 1214277 and tol 65 .
similar conclusions can be drawn for sbs 0335052 ( izotov et al .
1999 ) and sbs 0940 + 544 ( guseva et al .
2001 ) .
another approach has been developed by peimbert ( 1967 ) to take into account the difference in the electron temperature in the o@xmath51 zone as compared to the h@xmath46 and he@xmath46 zones .
he developed a formalism introducing an average temperature @xmath52 and a mean square temperature variation @xmath53 in an h ii region .
then the temperatures in the o@xmath51 and h@xmath46 and he@xmath46 zones are expressed as different functions of @xmath52 and @xmath53 , and in hot h ii regions @xmath54(o iii ) @xmath55 @xmath54(h ii ) , @xmath54(he ii ) .
this approach has been applied by peimbert et al .
( 2001 ) for the determination of the he abundance in some low - metallicity dwarf galaxies , including the two most - metal deficient bcds , i zw 18 and sbs 0335052 .
they use the observations by izotov et al .
( 1999 ) and find that while the ionization correction factors @xmath45(he ) in both galaxies are very close to unity , the difference in @xmath54(o iii ) and @xmath54(he ii ) results in the reduction of the he mass fraction by 2 3 percent compared to the case with @xmath54(o iii ) = @xmath54(he ii )
. additionally , peimbert et al . ( 2001 ) considered the effect of collisional excitation of hydrogen emission lines , first noted by davidson & kinman ( 1985 ) . neglecting this effect results in an artificially large extinction and hence overcorrection of the he i @xmath25876 and @xmath26678 emission lines .
peimbert et al .
( 2001 ) find that this effect in i zw 18 and sbs 0335052 leads to an the upward correction of @xmath4 by @xmath5 2% .
hence , after correction for all the systematic effects considered , they obtain @xmath4 = 0.241 and 0.245 , respectively for i zw 18 and sbs 0335052 .
these values are similar to @xmath4 = 0.243 and 0.246 derived earlier by izotov et al .
( 1999 ) for those galaxies .
the importance of the correction for collisional excitation of the hydrogen emission lines has been pointed out also by stasiska & izotov ( 2001 ) who concluded that this effect can result in an upward @xmath4 correction of up to 5% , assuming that the excess of the h@xmath32/h@xmath20 flux ratio above the theoretical recombination value is due only to collisional excitation . however , in practice , some part of the h@xmath32/h@xmath20 flux ratio excess is due to interstellar extinction and the correction of @xmath4 for collisional excitation of the hydrogen lines is likely smaller , only @xmath5 2 3 percent , similar to the value obtained by peimbert et al . (
2001 ) . because the physical conditions in the h ii regions of sbs 0940 + 544 ( guseva et al .
2001 ) , tol 1214277 and tol 65 ( this paper ) are similar to those in i zw 18 and sbs 0335052 , we expect that the estimates above for the systematic errors are valid for all the very metal - deficient high - excitation h ii regions considered in this paper . besides the above effects , the uncertainties in the he i recombination coefficients of @xmath5 1.5%
also may play a role ( benjamin , skillman & smits 1999 ) .
although some systematic effects are still difficult to estimate and are poorly studied , when taken into account together , it seems that they largely offset each other .
because of the uncertainties of these effects , we conservatively assume that combined systematic error in the he abundance determination may be tentatively set to 2% ( the error is 2@xmath6 ) .
our studies show that the helium mass fraction in the lowest - metallicity bcds observed with the keck telescope lies in the range 0.246 0.247 .
correction for the small contribution of @xmath3he produced in stars @xmath56@xmath4 = 0.0010 0.0017 ( izotov et al . 1999 ) , results in a mean primordial @xmath3he mass fraction @xmath13 of 0.245 @xmath0 0.003 ( rms ) @xmath0 0.005(sys ) ( 2@xmath6 ) obtained from the keck observations of the four galaxies , in agreement with the previous studies of itl97 , izotov & thuan ( 1998 ) and izotov et al .
this @xmath13 predicts a baryon mass fraction @xmath57 = [email protected](rms)@xmath58(sys ) ( 2@xmath6 ) , consistent with [email protected] ( 2@xmath6 ) derived from the primordial deuterium abundance ( burles & tytler 1998a , 1998b ; burles et al .
it is also consistent with the estimation of high-@xmath59 peaks in the angular power spectrum of the cosmic microwave background ( cmb ) ( netterfield et al . 2001 ) .
this overall consistency gives uniform support to the standard big bang nucleosynthesis model .
in particular , if the baryon mass fraction @xmath57 = [email protected] ( 1@xmath6 ) inferred from the cmb power spectrum is adopted , then , in the frame of the sbbn , the predicted primordial @xmath3he mass fraction is @xmath13 = [email protected] ( lopez & turner 1999 ; burles , nollett & turner 2001 ) .
our @xmath57 is also consistent with @xmath57 = [email protected] ( 1@xmath6 ) derived by pettini & bowen ( 2001 ) from the deuterium abundance measurements in the @xmath60 = 2.0762 damped lyman @xmath32 system toward the qso 2206199 if 2@xmath6 systematic error of 2% in @xmath13 is assumed .
the main conclusions drawn from our keck spectroscopic analysis of the extremely metal - deficient bcds tol 1214277 and tol 65 may be summarized as follows : \1 .
the oxygen abundances in tol 1214277 and tol 65 are 12 + log o / h = 7.54 @xmath0 0.01 , or 1/24 solar .
we find that the nitrogen - to - oxygen abundance ratio in both galaxies is log n / o = 1.64 @xmath0 0.02 , close to the mean value of 1.60 found for the other most - metal deficient bcds with @xmath61 @xmath36 @xmath1/20 ( thuan et al .
1995 ; izotov & thuan 1999 ) .
alpha - product element - to - oxygen abundance ratios are in the same range as those found for bcds .
the exception is the apparently higher fe / o abundance ratio in tol 1214277 which we argue is due to the contamination of [ fe iii ] @xmath24658 emission line by c iv @xmath24658 emission .
the @xmath3he mass fractions in tol 1214277 and tol 65 are respectively @xmath4 = [email protected] and [email protected] .
these values , after small corrections for the helium produced in stars , correspond to a primordial @xmath3he mass fraction of 0.245 , in excellent agreement with previous studies of itl97 , izotov & thuan ( 1998 ) and izotov et al .
( 1999 ) , supporting the validity of the standard big bang nucleosynthesis model .
we find that the systematic uncertainties in the @xmath3he abundance determination in tol 1214277 and tol 65 due to the combined effect of underlying stellar absorption , temperature and ionization structure of the h ii region and collisional excitation of the hydrogen emission lines are likely small , not exceeding @xmath5 2% ( 2@xmath6 ) .
aller , l. h. 1984 , physics of thermal gaseous nebulae ( dordrecht : reidel ) ballantyne , d. r. , ferland , g. j. , & martin , p. g. 2000 , , 536 , 773 benjamin , r. a. , skillman , e. d. , & smits , d. p. 1999
, , 514 , 307 bohlin , r. c. 1996 , , 111 , 1743 burles , s. , nollett , k. m. , & turner , m. s. 2001 , , 552 , l1 burles , s. , & tytler , d. 1998a , , 499 , 699 . 1998b , , 507 , 732 campbell , a. 1988 , , 335 , 644 campbell , a. , terlevich , r. , & melnick , j. 1986 , , 223 , 811 davidson , k. , & kinman , t. d. 1985 , , 58 , 321 filippenko , a. v. 1982 , , 94 , 715 fricke , k. j. , izotov , y. i. , papaderos , p. , guseva , n. g. , & thuan , t. x. 2001 , , 121 , 169 garnett , d. r. 1992 , , 103 , 1330 gonzlez delgado , r. , leitherer , c. , & heckman , t. m. 1999 , , 125 , 489 guseva , n. g. , izotov , y. i. , papaderos , p. , et al .
2001 , , in press izotov , y. i. , chaffee , f. h. , foltz , c. b. , green , r. f. , guseva , n. g. , & thuan , t. x. 1999 , , 527 , 757 izotov , y. i. , lipovetsky , v. a. , chaffee , f. h. , et al .
1997 , , 476 , 698 izotov , y. i. , schaerer , d. , & charbonnel , c. 2001 , , 549 , 878 izotov , y. i. , & thuan , t. x. 1998 , , 500 , 188 . 1999 , , 511 , 639 izotov , y. i. , thuan , t. x. , & lipovetsky , v. a. 1994 , , 435 , 647 ( itl94 ) . 1997c , , 108 , 1 ( itl97 ) kingdon , j. , & ferland , g. j. 1995 , , 442 , 714 kobulnicky , h. a. , & skillman , e. d. 1996 , , 471 , 211 kunth , d. , & sargent , w. l. w. 1983 , , 273 , 81 lopez , r. e. , & turner , m. s. 1999 , , 59 , 103502 masegosa , j. , moles , m. , & campos - aguilar , a. 1994 , , 420 , 576 mcgaugh , s. s. 1991 , , 380 , 140 netterfield , c. b. , et al .
2001 , , in press ; preprint astro - ph/0104460 oke , j. b. 1990 , , 99 , 1621 oke , j. b. , et al .
1995 , , 107 , 375 olive , k. a. , steigman , g. , & skillman , e. d. 1997 , , 483 , 788 pagel , b. e. j. 2000 , , 333 , 433 pagel , b. e. j. , simonson , e. a. , terlevich , r. j. , & edmunds , m. g. 1992 , , 255 , 325 pagel , b. e. j. , terlevich , r. j. , & melnick , j. 1986 , , 98 , 1005 peimbert , m. 1967 , , 150 , 825 peimbert , a. , peimbert , m. , & luridiana , v. 2001 , , in press ; preprint astro - ph/0107189 peimbert , m. , peimbert , a. , & ruiz , m. 2000 , , 541 , 688 peimbert , m. , & torres - peimbert , s. 1974 , , 193 , 327 . 1976 , , 203 , 581 pettini , m. , & bowen , d. v. 2001 , , in press ; preprint astro - ph/0104474 robbins , r. r. 1968 , , 151 , 511 sauer , d. , & jedamzik , k. 2001 , , in press ; preprint astro - ph/0104392 smits , d. p. 1996
, , 278 , 683 stasiska , g. 1990 , , 83 , 501 stasiska , g. , & izotov , y. i. 2001 , , in press steigman , g. , viegas , s. m. , & gruenwald , r. 1997 , , 490 , 187 thuan , t. x. , izotov , y. i. , & lipovetsky , v. a. 1995 , , 445 , 108 viegas , s. , gruenwald , r. , & steigman , g. 2000 , , 531 , 813 vlchez , j. m. , & pagel , b. e. j. 1988 , , 231 , 257 lccrcccr 3727 [ o ii ] & [email protected]&[email protected] & 42.3&&[email protected]&[email protected] & 67.1 + 3750 h12 & [email protected]&[email protected] & 4.1&&[email protected]&[email protected] & 1.3 + 3770 h11 & [email protected]&[email protected] & 4.5&&[email protected]&[email protected] & 2.7 + 3798 h10 & [email protected]&[email protected] & 6.8&&[email protected]&[email protected] & 3.6 + 3835 h9 & [email protected]&[email protected] & 7.1&&[email protected]&[email protected] & 4.9 + 3868 [ ne iii ] & [email protected]&[email protected] & 47.3&&[email protected]&[email protected] & 26.1 + 3889 he i + h8 & [email protected]&[email protected] & 28.2&&[email protected]&[email protected] & 16.2 + 3968 [ ne iii ] + h7 & [email protected]&[email protected] & 43.3&&[email protected]&[email protected] & 21.9 + 4026 he i & [email protected]&[email protected] & 3.0&&[email protected]&[email protected] & 1.3 + 4069 [ s ii ] & [email protected]&[email protected] & 1.2&&[email protected]&[email protected] & 1.2 + 4101 h@xmath62 & [email protected]&[email protected] & 45.6&&[email protected]&[email protected] & 25.8 + 4227 [ fe v ] & [email protected]&[email protected] & 1.3 & & & & + 4340 h@xmath63 & [email protected]&[email protected] & 96.9&&[email protected]&[email protected] & 58.0 + 4363 [ o iii ] & [email protected]&[email protected] & 34.0&&[email protected]&[email protected] & 12.3 + 4471 he i & [email protected]&[email protected] & 7.6&&[email protected]&[email protected] & 4.8 + 4658 [ fe iii ] & [email protected]&[email protected] & 1.2&&[email protected]&[email protected] & 1.0 + 4686 he
ii & [email protected]&[email protected] & 12.1&&[email protected]&[email protected] & 1.9 + 4711 [ ar iv ] + he i&[email protected]&[email protected] & 6.7&&[email protected]&[email protected] & 1.8 + 4740 [ ar iv ] & [email protected]&[email protected] & 4.0&&[email protected]&[email protected] & 0.9 + 4861 h@xmath20 & [email protected]&[email protected] & 267.5&&[email protected]&[email protected] & 174.1 + 4922 he i & [email protected]&[email protected] & 3.2&&[email protected]&[email protected] & 1.9 + 4959 [ o iii ] & [email protected]&[email protected] & 478.3&&[email protected]&[email protected] & 225.7 + 5007 [ o iii ] & [email protected]&[email protected] & 1467.2&&[email protected]&[email protected] & 693.4 + 5200
[ n i ] & & & & & [email protected]&[email protected] & 1.4 + 5876 he i & [email protected]&[email protected] & 41.4&&[email protected]&[email protected] & 31.4 + 6300 [ o i ] & [email protected]&[email protected] & 6.1&&[email protected]&[email protected] & 6.7 + 6312 [ s iii ] & [email protected]&[email protected] & 4.2&&[email protected]&[email protected] & 3.8 + 6363 [ o i ] & [email protected]&[email protected] & 1.5&&[email protected]&[email protected] & 2.6 + 6563 h@xmath32 & [email protected]&[email protected] & 1571.1&&[email protected]&[email protected] & 1079.6 + 6583 [ n ii ] & & & & & [email protected]&[email protected] & 4.7 + 6678 he i & [email protected]&[email protected] & 15.4&&[email protected]&[email protected] & 11.2 + 6717 [ s ii ] & [email protected]&[email protected] & 14.1&&[email protected]&[email protected] & 26.7 + 6731 [ s ii ] & [email protected]&[email protected] & 12.7&&[email protected]&[email protected] & 19.7 + 7065 he i & [email protected]&[email protected] & 17.1&&[email protected]&[email protected] & 13.6 + 7135 [ ar iii ] & [email protected]&[email protected] & 15.6&&[email protected]&[email protected] & 15.7 + 7281 he i & [email protected]&[email protected] & 4.0&&[email protected]&[email protected] & 2.9 + 7320 [ o ii ] & [email protected]&[email protected] & 3.8&&[email protected]&[email protected] & 7.6 + 7330 [ o ii ] & [email protected]&[email protected] & 3.0&&[email protected]&[email protected] & 5.4
+ + @xmath19(h@xmath20 ) dex & & & + @xmath21(h@xmath20 ) & & & + @xmath22(abs )
& & & + lccc @xmath23(o iii)(k ) & 19790@xmath0260 & & 17320@xmath0240 + @xmath23(o ii)(k ) & 15630@xmath0190 & & 14770@xmath0200 + @xmath23(s
iii)(k ) & 18130@xmath0210 & & 16080@xmath0200 + @xmath27(s ii)(@xmath64 ) & 400@xmath0120 & & 50@xmath050 + + o@xmath26/h@xmath26(@xmath1610@xmath65 ) & [email protected]&&[email protected] + o@xmath24/h@xmath26(@xmath1610@xmath65 ) & [email protected]&&[email protected] + o@xmath25/h@xmath26(@xmath1610@xmath65 ) & [email protected]&&[email protected] + o / h(@xmath1610@xmath65 ) & [email protected]&&[email protected] + 12
+ log(o / h ) & [email protected]&&[email protected] + + n@xmath46/h@xmath26(@xmath1610@xmath8 ) & & & [email protected] + icf(n ) & & & 5.66 + log(n / o ) & & & [email protected] + + ne@xmath24/h@xmath26(@xmath1610@xmath65 ) & [email protected]&&[email protected] + icf(ne ) & 1.16 & & 1.23 + log(ne / o ) & [email protected] & & [email protected] + + s@xmath26/h@xmath26(@xmath1610@xmath8 ) &
[email protected]&&[email protected] + s@xmath24/h@xmath26(@xmath1610@xmath8 ) & [email protected]&&[email protected] + icf(s ) & 2.90 & & 1.69 + log(s / o ) & [email protected] & & [email protected] + + ar@xmath24/h@xmath26(@xmath1610@xmath8 ) & [email protected]&&[email protected] + ar@xmath25/h@xmath26(@xmath1610@xmath8 ) & [email protected]&&[email protected] + icf(ar ) & 1.01 & & 1.03 + log(ar / o ) & [email protected] & & [email protected] + + fe@xmath24/h@xmath26(@xmath1610@xmath8 ) & [email protected]&&[email protected] + icf(fe ) & 15.8 & & 7.07 + log(fe / o ) & [email protected] & & [email protected] + @xmath66o / fe@xmath67 $ ] & [email protected] & & [email protected] + lccc @xmath23(o iii)(k ) & 19790@xmath0260 & & 17320@xmath0240 + @xmath27(he
ii)(@xmath64 ) & 25@xmath01 & & 150@xmath050 + @xmath39(@xmath23889 ) & 0.01 & & 0.01
+ + @xmath38(@xmath24471 ) & [email protected]&&[email protected] + @xmath38(@xmath25876 ) & [email protected]&&[email protected] + @xmath38(@xmath26678 ) & [email protected]&&[email protected]
+ @xmath38(weighted mean ) & [email protected]&&[email protected] + @xmath38(@xmath25876 + @xmath26678)&[email protected]&&[email protected] + @xmath40(@xmath24686 ) & [email protected]&&[email protected] + + @xmath42(@xmath24471 ) & [email protected]&&[email protected] + @xmath42(@xmath25876 ) & [email protected]&&[email protected] + @xmath42(@xmath26678 ) & [email protected]&&[email protected] + @xmath42(weighted mean ) & [email protected]&&[email protected] + @xmath42(@xmath25876 + @xmath26678 ) & [email protected]&&[email protected] + + @xmath4(@xmath24471 ) & [email protected]&&[email protected] + @xmath4(@xmath25876 ) & [email protected]&&[email protected] + @xmath4(@xmath26678 ) & [email protected]&&[email protected] + @xmath4 ( weighted mean ) & [email protected]&&[email protected] + @xmath4 ( @xmath25876 + @xmath26678 ) & [email protected]&&[email protected] + | we present high - quality keck telescope spectroscopic observations of the two metal - deficient blue compact dwarf ( bcd ) galaxies tol 1214277 and tol 65 .
these data are used to derive the heavy - element and helium abundances .
we find that the oxygen abundances in tol 1214277 and tol 65 are the same , 12 + log o / h = 7.54 @xmath0 0.01 , or @xmath1/24 , despite the different ionization conditions in these galaxies .
the nitrogen - to - oxygen abundance ratio in both galaxies is log n / o = [email protected] and lies in the narrow range found for the other most metal - deficient bcds .
we use the five strongest he i emission lines @xmath2@xmath23889 , 4471 , 5876 , 6678 and 7065 , to correct self - consistently their intensities for collisional and fluorescent enhancement mechanisms and to derive the @xmath3he abundance .
underlying stellar absorption is found to be important for the he i @xmath24471 emission line in both galaxies , being larger in tol 65 .
the weighted @xmath3he mass fractions in tol 1214277 and tol 65 are respectively @xmath4 = 0.2458 @xmath0 0.0039 and 0.2410 @xmath0 0.0050 when the three he i emission lines , @xmath2@xmath24471 , 5876 and 6678 , are used , and are , respectively , 0.2466 @xmath0 0.0043 and 0.2463 @xmath0 0.0057 when the he i 4471 emission line is excluded .
these values are in very good agreement with recent measurements of the @xmath3he mass fraction in others of the most metal - deficient bcds by izotov and coworkers .
we find that the combined effect of the systematic uncertainties due to the underlying he i stellar absorption lines , ionization and temperature structure of the h ii region and collisional excitation of the hydrogen emission lines is likely small , not exceeding @xmath5 2% ( the error is 2@xmath6 ) .
our results support the validity of the standard big bang model of nucleosynthesis . |
let @xmath3 be a smooth , connected , projective complex curve of genus @xmath12 .
let @xmath11 be the hurwitz space which parametrizes degree @xmath2 coverings of @xmath3 simply branched in @xmath5 points . a classical result of hurwitz @xcite states that @xmath13 is irreducible .
more generally one can consider coverings of @xmath14 which are simply branched in all but one point of the discriminant .
fixing the ramification multiplicities over the special point one obtains a corresponding hurwitz space which turns out to be irreducible as well ( see @xcite , @xcite , @xcite ) .
coverings of curves of positive genus were studied by graber , harris and starr in @xcite .
they considered the hurwitz space @xmath1 which parametrizes coverings with full monodromy group @xmath6 and proved it is irreducible if @xmath0 .
another result of this type was obtained by f. vetro @xcite . in the present paper
we sharpen the result of graber , harris and starr and prove that @xmath1 is irreducible if @xmath15 ( cf . theorem [ s4.45 ] ) .
our approach allows to extend the result to coverings which are simply branched in all but one point of the discriminant .
fixing the branching data of the special point , i.e. a partition @xmath16 where @xmath17 and @xmath18 one obtains a hurwitz space @xmath19 parametrizing coverings with full monodromy group , simply branched in @xmath5 points and ramified with multiplicities @xmath20 over one additional point .
we prove @xmath19 is irreducible if @xmath21 ( cf .
theorem [ s4.49 ] ) .
coverings with @xmath22 are more difficult to deal with .
we prove the irreducibility of @xmath1 when @xmath23 ( cf .
theorem [ s5.64bis ] ) and the irreducibility of @xmath1 when @xmath24 ( cf .
theorem [ s6.81 ] ) .
the case of simply branched coverings with monodromy groups @xmath10 can be reduced via tale coverings to the case of coverings with full monodromy group . in theorem [ s4.55 ] and
theorem [ s6.87 ] we prove that connected components of @xmath11 not contained in @xmath1 may exist only if @xmath2 is not prime and in this case if @xmath5 is sufficiently large such connected components correspond bijectively to the equivalence classes of tale coverings @xmath25 $ ] of degree @xmath26 , where @xmath27 .
in proving these results we follow the standard approach for determining the connected components of @xmath11 .
associating to every equivalence class of coverings @xmath28\in \mathcal{h}_{d , n}(y)$ ] its discriminant locus yields an tale mapping @xmath29 . fixing a @xmath30
the fiber over @xmath31 is identified via riemann s existence theorem to the equivalence classes modulo inner automorphisms of ordered @xmath32-tuples @xmath33 satisfying @xmath34\cdots [ \lambda_{g},\mu_{g}]$ ] we call such @xmath32-tuples _ hurwitz systems _ where @xmath35 , @xmath36 are transpositions , and @xmath37 with @xmath38 generate a transitive subgroup of @xmath6 .
the problem is thus reduced to finding the orbits of the action of the @xmath5-strand braid group of @xmath3 , namely @xmath39 , on the set of equivalence classes of hurwitz systems .
in fact it is more natural and it suffices to study the action of the braid group of the open riemann surface @xmath40 , where @xmath41 is a fixed point .
birman found in @xcite a natural system of generators of these braid groups .
our results are based on theorem [ s2.27 ] where we calculate explicitly how these generators act on hurwitz systems .
we follow the key idea of @xcite which consists in applying a sequence of braid moves to a given hurwitz system @xmath33 in order to replace it by a new one @xmath42 in which as many as possible of the elements @xmath43 equal 1 and then reduce the sequence @xmath44 to a normal form using only elementary transformations of the artin s braid group .
an important tool in our arguments is main lemma [ s3.32 ] which states that if in a hurwitz system @xmath33 one has @xmath45 , then replacing the pair @xmath46 by a pair @xmath47 , where @xmath48 belongs to the subgroup generated by @xmath49 with @xmath50 , one obtains a braid - equivalent hurwitz system , i.e. can be obtained from the initial one by a finite sequence of braid moves .
this is a generalization of lemma 2.2 of @xcite proved there using the kontsevich moduli space of stable maps .
the proof of our main lemma [ s3.32 ] is more elementary and uses only the explicit formulas for the braid moves of theorem [ s2.27 ] .
we think theorem [ s2.27 ] and main lemma [ s3.32 ] are results of independent interest , since both are valid for hurwitz systems with values in an arbitrary ( possibly infinite ) group @xmath51 .
we profitted a lot from the paper of mochizuki @xcite which was the starting point for us in studying the problem of the irreducibility of hurwitz spaces .
we assume that the base field @xmath52 . throughout the paper @xmath53
denotes a covering , i.e. a finite holomorphic mapping , of smooth projective curves .
we do not assume that @xmath54 is connected , while @xmath3 is always assumed connected .
two coverings @xmath55 and @xmath56 are called equivalent if there is an isomorphism @xmath57 such that @xmath58 .
a covering @xmath53 of degree @xmath2 is called simple if both @xmath54 and @xmath3 are irreducible and for each @xmath59 one has @xmath60 .
we use mainly right actions and write @xmath61 for @xmath62 and @xmath63 , where the group @xmath51 acts on the set @xmath64 on the right . in particular if @xmath65 we let @xmath66 . \1 .
braid moves , 2 .
the main lemma , 3 .
the case @xmath21 , 4 .
the case @xmath67 , 5 .
the case @xmath68 .
[ s2.11 ] the equivalence classes of ( possibly nonconnected ) coverings of degree @xmath2 of a smooth , projective , irreducible curve @xmath3 branched in @xmath69 points are parametrized by a smooth scheme @xmath70 which is an tale cover of @xmath71 . here
@xmath72 is the codimension one subscheme consisting of nonsimple divisors of degree @xmath5 .
the covering mapping associates to @xmath28 $ ] its branch locus @xmath31 .
let @xmath73 .
we denote by @xmath74 the open subset of @xmath70 consisting of coverings which are unramified at @xmath75 .
similarly we consider equivalence classes of pairs @xmath76 $ ] where @xmath28\in \mathcal{u}(b_0)$ ] and @xmath77 is a bijection .
we denote by @xmath78 the corresponding tale cover of @xmath74 .
let @xmath79 .
riemann s existence theorem ( see e.g. @xcite p.544 ) establishes the following natural one - to - one correspondences 1 . for @xmath78 between : 1 .
the fiber of @xmath80 over @xmath31 2 .
the set of homomorphisms @xmath81 which satisfy the condition that @xmath82 for each small loop which circles a point of @xmath31 .
2 . for @xmath70 between : 1 . the fiber of @xmath83 over @xmath31 2 . the set of equivalence classes modulo inner automorphisms of homomorphisms @xmath84 satisfying the condition of a(ii ) , here @xmath85 if there exists an @xmath86 such that @xmath87 .
we fix our preferences in this paper as follows .
we consider the product in a fundamental group induced by product of arcs defined as @xmath88 we consider right action of @xmath6 on @xmath89 , so a permutation @xmath90 transforms @xmath91 into @xmath92 and for instance @xmath93 . explicitly the correspondence in ( a ) is defined in the following way . given a pair @xmath94 let us number the elements of the fiber @xmath95 using @xmath96 , thus @xmath97 where @xmath98 .
then given a homotopy class of loops @xmath99 $ ] with @xmath100 the permutation @xmath101 acts as follows .
if lifting @xmath102 one starts from @xmath103 and ends at @xmath104 , then one lets @xmath105 .
let @xmath86 .
renumbering the points of the fiber @xmath95 by a bijective map @xmath106 such that @xmath107 results in replacing the monodromy homomorphism @xmath108 by @xmath109 . hence the one - to - one correspondence in ( b ) is obtained from that of ( a ) applying the forgetful mapping @xmath110\mapsto [ x\to y]$ ] . [ s2.12 ] suppose that @xmath111 .
let us fix the orientation of the real 2-manifold @xmath3 considered as a complex manifold .
let @xmath112 .
let @xmath113 .
we describe a standard way of choosing generators of @xmath114 . cutting @xmath3 along @xmath115 simple closed arcs which begin at @xmath41 and do not contain any of @xmath116 one obtains a standard @xmath117-polygon with sides @xmath118 which circle the polygon in counterclockwise direction .
we consider a simple closed arc @xmath119 which begins at @xmath75,@xmath120 is contained in the interior of the 4g - polygon and passes consecutively in counterclockwise direction through the points @xmath121 .
the closed arc @xmath119 divides the @xmath117 polygon into two regions @xmath122 and @xmath123 which stay on the left respectively on the right side of @xmath119 with respect to its counterclockwise orientation .
we choose a simple arc @xmath124 which lies inside the region @xmath122 and connects @xmath75 and @xmath125 .
then we choose a second simple arc @xmath126 inside @xmath122 which connects @xmath75 with @xmath127 , has only @xmath75 as point in common with @xmath124 , and lies on the left side of @xmath124 .
continuing in this way we obtain an ordered @xmath5-tuple @xmath128 of simple arcs which do not meet outside @xmath75 .
let @xmath129 be a closed path which begins at @xmath75 , travels along @xmath130 to a point near @xmath131 , makes a small counterclockwise loop around @xmath131 , and returns to @xmath75 along @xmath130 .
we obtain a @xmath32-tuple of closed arcs @xmath132 which we call a _ standard system of closed arcs_. the corresponding homotopy classes yield a standard system of generators for @xmath114 which satisfy the only relation @xmath133\cdots [ \alpha_{g},\beta_{g}]\ ] ] figure [ fs2.13a ] illustrates such a standard system . a reader who prefers the clockwise orientation of closed arcs and ordering of the branch points from left to right
may look at this and all subsequent figures from the other side of the sheet . given a covering @xmath134 with discriminant @xmath31 and an isomorphism @xmath135 one applies the monodromy homomorphism @xmath108 and obtains @xmath136 .
[ s2.14 ] an ordered sequence @xmath137 of permutations in @xmath6 with @xmath138 for @xmath139 and satisfying the relation @xmath140\cdots [ \lambda_{g},\mu_{g}]\ ] ] is called a _ hurwitz system_. we let @xmath141 .
we call the subgroup @xmath142 generated by all @xmath143 the _ monodromy group _ of the hurwitz system . by [ s2.11 ] given @xmath144 and @xmath145 , and fixing the closed arcs @xmath146 as above , the fiber of @xmath147 over @xmath31 may be identified with the set of all hurwitz systems , while the fiber of @xmath148 over @xmath31 may be identified with the set of equivalence classes @xmath149 $ ] of hurwitz systems modulo inner automorphisms of @xmath6 , where @xmath150 is equivalent to @xmath151 if there exists an @xmath86 such that @xmath152 for @xmath153 .
an equivalent way of constructing a standard system of closed arcs is the following .
one chooses first the @xmath115 simple closed arcs @xmath154 .
then one chooses @xmath5 simple arcs which start at @xmath41 , lie inside the @xmath117-polygon , do not meet outside @xmath41 , and have for end points the @xmath5 points of @xmath31 .
one enumerates these arcs according to the directions of departure in counterclockwise order .
the obtained @xmath32-tuple @xmath155 is called an _ arc system _ ( cf .
@xcite p.416 ) .
one considers the induced ordering of the points of @xmath31 .
one can take for @xmath122 a star - like region which contains the union @xmath156 and let @xmath157 . in this way one obtains all ingredients used to construct a standard system of closed arcs @xmath132 .
[ s2.14a ] the connected components of @xmath70 are in one - to - one correspondence with the orbits of the full @xmath5-strand braid group @xmath39 acting on the fiber of the topological covering @xmath148 over @xmath31 .
similar statement holds about @xmath78 where one considers the braid group @xmath158 .
the identification of these fibers with the hurwitz systems reduces the problem of determining the connected components of @xmath78 and @xmath70 to calculating the action of the respective braid groups on hurwitz systems and then finding the orbits .
let @xmath159 be a standard system of closed arcs obtained from an arc system @xmath160 as in [ s2.12 ] .
let @xmath161 be a closed arc in @xmath162 with @xmath163 .
suppose that starting from the given arc system one can extend the map @xmath164 to a homotopy of arc systems @xmath165 based at @xmath41 .
this yields a corresponding homotopy of closed arcs @xmath166 which form a standard system for each @xmath167 $ ] .
let @xmath168\in \mathcal{h}_d^{(n)}(y , b_0)$ ] have discriminant @xmath31 and let @xmath169 be the corresponding hurwitz system .
then the lifting of the closed arc @xmath170 , starting from a @xmath171 $ ] , is defined by by @xmath172 where ( cf . @xcite p.545 ) @xmath173 letting @xmath174 and @xmath175 we obtain the end point of the lifting of @xmath176 is @xmath177 $ ] whose monodromy map @xmath178 is defined by @xmath179 evaluating @xmath180 at @xmath181 we obtain @xmath182 so @xmath183 is the hurwitz system corresponding to @xmath177 $ ] .
another approach is to consider @xmath184 the hurwitz system @xmath185 corresponds to a pair @xmath186 $ ] whose monodromy map @xmath187 satisfies @xmath188 [ s2.16 ] let us lift the path @xmath189 starting from @xmath190 $ ] .
then the end point is @xmath186 $ ] . in terms of @xmath191 the monodromy map of @xmath171
$ ] is given by @xmath192 .
let us consider the homotopy of closed paths @xmath193 , @xmath194 and the path in @xmath78 given by @xmath195 where @xmath196
. then @xmath197 ( cf
. eq . )
[ s2.17 ] given a closed arc @xmath170 in the configuration space @xmath198 the transformation of hurwitz systems @xmath199 given by eq .
is called a _ braid move _ of the first type . the transformation @xmath200 given by eq .
is called a braid move of the second type . the braid moves of the first and of the second type are inverse to each other according to lemma [ s2.16 ] .
it is evident that the braid moves of both types commute with inner automorphisms of @xmath201 .
so the braid moves are well - defined on equivalence classes of hurwitz systems ( cf .
[ s2.12 ] ) .
[ s2.17a ] there is a convenient system of generators of @xmath158 .
we include here some material borrowed from @xcite and @xcite for the sake of convenience of the reader and since our choices differ slightly from theirs .
consider the galois covering @xmath202 with galois group @xmath203 .
restricting to the complement of @xmath72 one obtains an unramified galois covering @xmath204 .
following the notation of @xcite if @xmath205 one denotes @xmath206 by @xmath207 .
let @xmath208 .
one has an exact sequence @xmath209 one determines first a system of generators of the pure braid group @xmath210 as follows .
consider the closed arcs in @xmath207 defined by @xmath211 and @xmath212 , with @xmath213 $ ] , where @xmath214 and @xmath215 are the closed simple arcs based at @xmath131 and pictured on figure [ fs2.17a ] by a continuous line and by a dotted line respectively . ]
we denote the corresponding homotopy classes by @xmath216 .
informally @xmath217 corresponds to a loop of the @xmath91-th point along @xmath218 and @xmath219 corresponds to a loop of the @xmath91-th point along @xmath220 .
let us denote by @xmath221 the element of @xmath210 represented by a closed simple arc in @xmath222 which leaves fixed @xmath223 for @xmath224 and moves @xmath131 along the arc pictured on figure [ fs2.17a ] .
for every @xmath91 and every @xmath225 the pictured loop is chosen so as to stay on the left of all arcs used to construct @xmath217 and @xmath219 for @xmath226 .
the pure braid group @xmath210 is generated by @xmath227 and @xmath228 where @xmath229 and @xmath226 .
this is proved by induction on @xmath5 . when @xmath230 the claim is obvious .
let @xmath231 .
consider the fibration @xmath232 defined by @xmath233 .
one has an exact sequence @xmath234 since @xmath235 by @xcite , corollary 2.2 .
the elements @xmath236 with @xmath237 map to the corresponding elements in @xmath238 . the group @xmath239 is freely generated by elements which map into @xmath240 where @xmath241 .
this shows the claim .
following @xcite p.547 and using the notation of [ s2.12 ] let us denote by @xmath242 the simply connected region of @xmath122 enclosed by the arcs @xmath130 and @xmath243 and the arc of @xmath119 from @xmath244 to @xmath245 .
for every @xmath246 choose simple arcs @xmath247\to y$ ] going from @xmath131 to @xmath245 in @xmath242 and @xmath248\to y$ ] going from @xmath245 to @xmath131 in @xmath123 .
let @xmath249\to y^{(n)}-\delta$ ] be the closed arc @xmath250 the homotopy class of @xmath251 is denoted by @xmath252 .
we may consider @xmath253 as embedded in @xmath158 .
the following relations are easy to verify ( cf .
@xcite p.249 ) @xmath254 [ s2.19 ] let @xmath255 .
the braid group @xmath158 is generated by @xmath256 where @xmath257 .
the corresponding homotopy classes generate @xmath39 as well .
let us consider the exact sequence eq .. the braids @xmath258 map to the transpositions @xmath259 which generate @xmath203 .
so @xmath260 with @xmath261 generate @xmath158 according to the claim proved in [ s2.17a ] . the relations eq .
show that the generators may be reduced as stated in the theorem .
the last statement follows from the surjection @xmath262 .
we described in [ s2.17a ] closed arcs in @xmath198 based at @xmath263 whose homotopy classes form a system of generators @xmath264 for @xmath158 .
our aim now is for each of these to construct a homotopy of the standard system of closed arcs @xmath265 as in [ s2.14a ] .
this will permit us to calculate eventually the corresponding braid moves of the hurwitz systems .
the calculation of the braid moves @xmath266 is due to hurwitz ( cf .
@xcite or e.g. @xcite , theorem 10.3 ) .
we define closed arcs @xmath267\to y,\ : \delta_k(0)=\delta_k(1)=b_{0},\ : k=0,1,\ldots , g$ ] as follows .
we let @xmath268 $ ] . we connect the initial vertex of @xmath269 with the end vertex of @xmath270 in the @xmath117-polygon of figure [ fs2.13a ] by a simple arc which belongs to the region @xmath123 on the right of @xmath119 ( cf .
[ s2.12 ] ) .
this yields @xmath271 .
we connect the initial vertex of @xmath269 with the end vertex of @xmath272 by a simple arc which belongs to the region on the right of @xmath119 and on the left of @xmath271 .
we denote the corresponding closed arc of @xmath3 by @xmath273 . continuing in this way
we obtain @xmath274 ( see figure [ fs2.22 ] ) .
clearly @xmath275\cdots [ \alpha_{k},\beta_{k}]$ ] in @xmath276 and @xmath277 according to eq .. [ s2.20 ] let @xmath3 be a compact , closed riemann surface of genus @xmath111 .
let @xmath73 and let @xmath278 with @xmath279 be a standard system of closed arcs as in [ s2.12 ] .
let @xmath280 be the closed arcs defined above .
for each of the closed arcs in @xmath198 constructed in [ s2.17a ] and representing @xmath281 there is a homotopy @xmath282 $ ] of the standard system of closed arcs such that the end system @xmath283 is homotopic to : a.for @xmath284 where @xmath285 @xmath286 b.for @xmath287 where @xmath288 @xmath289 c.for @xmath290 where @xmath288 @xmath291 for each of the inverse closed arcs corresponding to @xmath292 there is a corresponding homotopy of the standard system @xmath278 , @xmath288 such that the end system @xmath293 , @xmath288 is homotopic to : d.for @xmath294 where @xmath285 @xmath295 e.for @xmath296 where @xmath288 @xmath297 f.for @xmath298 where @xmath288 @xmath299 ( a)here one moves only @xmath300 and @xmath301 .
clearly @xmath302 and @xmath303 , so @xmath304 .
( b)the effect of the homotopy of @xmath278 along @xmath287 is pictured on figure [ fs2.22 ] .
one moves the point @xmath131 along the arc @xmath214 pictured on figure [ fs2.17a ] and together with it deforms the closed arc @xmath129 . at the moment @xmath305 reaches the side @xmath220 one deforms also @xmath220 in order that the condition @xmath305 and @xmath306 have no points in common except @xmath75 remains valid .
none of @xmath300 for @xmath307 , or @xmath308 for @xmath309 , or @xmath310 for @xmath311 changes in this homotopy .
the effect of cutting @xmath3 along the closed arcs @xmath312 is the same as to cut a region containing @xmath131 from the original @xmath117-polygon and glue it along the side @xmath313 as described in figure [ fs2.22 ] ( nw ) ( nw = northwest ) .
we wish to express @xmath314 in terms of the standard system @xmath315 , @xmath316 .
the closed arc @xmath314 is pictured in the original @xmath117-polygon on figure [ fs2.22 ] ( ne ) .
it is clear that @xmath317 where @xmath318 is the closed arc pictured on figure [ fs2.22 ] ( sw ) . that @xmath318 is homotopic to @xmath319 is evident from figure [ fs2.22 ] ( se ) .
furthermore @xmath320 . using the relation @xmath321
we obtain @xmath322 .
this proves the formula for @xmath314 of part ( b ) .
the calculation of @xmath323 is similar .
the closed arc @xmath323 is pictured in the original @xmath117-polygon on figure [ fs2.25 ] ( nw ) .
it is homotopic to @xmath324 where @xmath325 is the closed arc based at @xmath75 pictured on figure [ fs2.25 ] ( ne ) .
we then consider @xmath326 ( see figure [ fs2.25 ] ( sw ) ) .
the latter is homotopic to @xmath327 .
we thus obtain @xmath328 .
we then have @xmath329\gamma_{i}[\gamma_{i+1}\cdots \gamma_{n } \delta_{g}^{-1}\delta_{k}]\end{aligned}\ ] ] since @xmath330 .
this proves the second formula of part ( b ) .
( c)the arguments here are very similar to those of part ( b ) .
one deforms @xmath129 and @xmath218 with @xmath331 moving along the arc @xmath215 ( see figure [ fs2.17a ] ) .
when @xmath331 returns to @xmath131 one obtains closed arcs @xmath332 and @xmath333 for which @xmath334 , @xmath335 where @xmath336 and @xmath337 are represented by arcs in the original @xmath117-polygon pictured on figure [ fs2.26 ] .
we calculate @xmath336 in the following way .
we consider it as a product of four arcs according to the picture .
we deform @xmath336 in such a way that the first arc becomes a closed simple arc encircling @xmath338 in clockwise direction , then the second arc goes from the initial point of @xmath269 to the end point of @xmath339 , the third arc goes from the initial point of @xmath218 to the initial point of @xmath269 and the fourth arc equals the first one with the opposite orientation .
we have accordingly @xmath340 conjugating @xmath129 by @xmath336 we may cancel the last factor @xmath129 from @xmath336 and replace @xmath341 by @xmath342 .
we thus obtain @xmath343 where @xmath344 finally , @xmath345 .
this proves the last formula of part ( c ) . in order to obtain the formulas of ( d ) , ( e ) and ( f ) from those of ( a ) , ( b ) and ( c ) we notice that if we apply in each case to @xmath346 the braid moves @xmath347 respectively we obtain @xmath348 .
for example , in order to verify ( d ) we have @xmath349
. therefore @xmath350 . in case ( e ) applying @xmath287 to @xmath351 we obtain @xmath352 for @xmath353 for @xmath309 , @xmath354 for @xmath355 .
furthermore @xmath356 where @xmath357 is expressed by @xmath358 as in ( b ) .
since neither @xmath359 nor @xmath360 enter in this expression we may replace @xmath358 by @xmath315 and we obtain @xmath361 . thus @xmath362 .
similarly we have @xmath363 where @xmath364 is as in ( b ) . replacing @xmath359 by @xmath365 and canceling we
obtain @xmath366 where @xmath367 is as in ( e ) . in a similar manner one deduces ( f ) from ( c ) .
recall from [ s2.11 ] and [ s2.12 ] that given @xmath368 and fixing a standard system of closed arcs @xmath369 there is a bijective correspondence between the fiber of @xmath147 over @xmath31 and the set of hurwitz systems @xmath33 .
similarly there is a bijective correspondence between the fiber of @xmath370 , where @xmath371 , and the set of equivalence classes of hurwitz systems @xmath372 $ ] . in the next theorem
we calculate the monodromy action of @xmath158 on these fibers . according to proposition [ s2.19 ] it suffices to determine the braid moves which correspond to the generators @xmath347 .
we denote the corresponding braid moves of the first type ( cf .
definition [ s2.17 ] ) by @xmath373 and we denote the corresponding braid moves of the second type ( inverse to those of the first type ) by @xmath374 . [ s2.27 ] let @xmath33 , @xmath141 , be a hurwitz system .
let @xmath375\cdots [ \lambda_{k},\mu_{k}]$ ] for @xmath376 and let @xmath377 .
the following formulas hold for the braid moves @xmath378 of the first type .
a.for @xmath379 where @xmath285 @xmath380 b.for @xmath381 where @xmath382 @xmath383 c.for @xmath384 where @xmath382 @xmath385 the following formulas hold for the braid moves @xmath386 of the second type .
d.for @xmath387 where @xmath285 @xmath388 e.for @xmath389 where @xmath382 @xmath390 f.for @xmath391 where @xmath382 @xmath392 formulas ( d ) , ( e ) and ( f ) are obtained from formulas ( a ) , ( b ) and ( c ) of theorem [ s2.20 ] respectively applying the homomorphism @xmath393 and equalities eq . and
eq .. by lemma [ s2.16 ] the braid moves of the first type are inverse to the braid moves of the second type .
we may thus obtain formulas ( a ) , ( b ) and ( c ) from formulas ( d ) , ( e ) and ( f ) of theorem [ s2.20 ] respectively applying the homomorphism @xmath108 .
[ s2.30a ] using the notation of theorem [ s2.27 ] the following formulas hold . 1 . for @xmath381 : @xmath394\lambda_{k } ( u_{k}^{-1}u_{g})$}\\ & \rho_{1k}':\mu_{k}\mapsto \mu_{k}'=(b_{1}^{-1}t_{1}^{-1}b_{1})\mu_{k } , & & \text{where $ b_{1}=u_{k-1}\lambda_{k}$}\\ & \rho_{nk}':\mu_{k}\mapsto \mu'_{k}=(b_{1}^{-1}t_{n}^{-1}b_{1})\mu_{k } , & & \text{where $ b_{1}= ( u_{k}^{-1}u_{g})^{-1}[\mu_{k},\lambda_{k}]\lambda_{k}$}.\end{aligned}\ ] ] 2 . for @xmath395 : @xmath396\mu_{k}u_{k-1}^{-1}$\hspace{0.3cm}}\\ & \tau'_{1k}:\lambda_{k}\mapsto \lambda_{k}'=(d_{1}^{-1}t_{1}d_{1})\lambda_{k } , & & \text{where $ d_{1}=u_{k-1}[\lambda_{k},\mu_{k}]\mu_{k}$}\\ & \tau_{nk}':\lambda_{k}\mapsto \lambda_{k}'= ( d_{1}^{-1}t_{n}d_{1})\lambda_{k } , & & \text{where $ d_{1}=(u_{k}^{-1}u_{g})^{-1}\mu_{k}$}.\end{aligned}\ ] ] 3 . for @xmath397 : @xmath398\lambda_{k}^{-1 } u_{k-1}^{-1}$}\\ & \rho_{1k}'':\mu_{k}\mapsto \mu_{k}''=(u_{k}^{-1}t_{1}u_{k})\mu_{k } , & & \\ & \rho_{nk}'':\mu_{k}\mapsto \mu''_{k}=(b_{1}^{-1}t_{n}b_{1})\mu_{k } , & & \text{where $ b_{1}= ( u_{k}^{-1}u_{g})^{-1}$}.\end{aligned}\ ] ] 4 . for @xmath399 : @xmath400\mu_{k}^{-1 } ( u_{k}^{-1}u_{g})$}\\ & \tau''_{1k}:\lambda_{k}\mapsto \lambda_{k}''=(u_{k-1}^{-1}t_{1}^{-1}u_{k-1})\lambda_{k } , & & \\ & \tau_{nk}'':\lambda_{k}\mapsto \lambda_{k}''= ( d_{2}^{-1}t_{n}^{-1}d_{2})\lambda_{k } , & & \text{where $ d_{2}=(u_{k}^{-1}u_{g})^{-1}[\mu_{k},\lambda_k]$}. \ ] ] in particular @xmath401 let us prove the first formula of part ( ii ) .
the other formulas can be either proved similarly or are restatements of particular cases of theorem [ s2.27 ] .
we have @xmath402 where @xmath403 . since @xmath404 it holds @xmath405 . hence @xmath406^{-1 } u_{k-1}^{-1})t_1(\cdots)^{-1}\\ & = \ ( u_{k-1}[\lambda_{k},\mu_{k}]\mu_{k}u_{k-1}^{-1})^{^{-1}}\ , t_1\ , ( u_{k-1}[\lambda_{k},\mu_{k}]\mu_{k}u_{k-1}^{-1 } ) \end{split}\ ] ] summing up the discussion made so far in this section we obtain the following result . [ s2.30c ] fix @xmath407 .
let us consider the set of all hurwitz systems @xmath33 ( cf . definition [ s2.14 ] ) .
let @xmath408 be the free group generated by the symbols @xmath347 where @xmath409 .
let us consider the right action of @xmath408 on the set of hurwitz systems defined by the formulas for @xmath373 of theorem [ s2.27 ] ( a)(c ) or alternatively let us consider the left action of @xmath408 on the set of hurwitz systems defined by the formulas for @xmath374 of theorem [ s2.27 ] ( d)(f )
. then the connected components of @xmath78 ( cf .
[ s2.11 ] ) correspond bijectively to the orbits of either action of @xmath408 and the connected components of @xmath70 correspond bijectively to the orbits of either associated action of @xmath408 on the set of equivalence classes of hurwitz systems modulo inner automorphisms of @xmath6 .
@xmath147 is a topological covering map where a hausdorff topology on @xmath78 is defined as in @xcite pp.545,546 . according to the definition of product of arcs
the monodromy action of the fundamental group @xmath158 on the fiber over @xmath31 is a right action .
the identification of this fiber with the set of hurwitz systems , proposition [ s2.19 ] and the calculation of the braid moves of the first type @xmath373 in theorem [ s2.27 ] yield the statement about the orbits of the right action of @xmath408 .
consider the associated left action @xmath410 .
the orbits are the same and according to lemma [ s2.16 ] the braid moves of the second type @xmath374 are inverse to those of the first type @xmath373 respectively .
this shows the statement about the orbits of the left action of @xmath408 .
let @xmath73 . since @xmath411 ( cf .
[ s2.11 ] ) is a zariski dense open subset , the connected components of @xmath412 correspond bijectively to those of @xmath70 .
this shows the last claim of the theorem .
in the previous section we considered hurwitz systems @xmath33 where @xmath413 in connection with the problem of determining the connected components of the hurwitz spaces . in this section
we replace @xmath6 by an arbitrary ( possibly infinite ) group @xmath51 .
[ s3.31a ] let @xmath51 be an arbitrary group .
we call @xmath33 a hurwitz system with values in @xmath51 if @xmath414 , @xmath415 for @xmath139 and @xmath34\cdots [ \lambda_{g},\mu_{g}]$ ]
. the formulas of theorem [ s2.27 ] make sense for an arbitrary group @xmath51 . that @xmath416 and @xmath417 is evident from theorem [ s2.20 ] .
[ s3.31ab ] we call two hurwitz systems with values in @xmath51 braid - equivalent if one is obtained from the other by a finite sequence of braid moves @xmath418 where @xmath419 .
we denote the braid equivalence by @xmath420 .
[ s3.32 ] let @xmath51 be an arbitrary group and let @xmath33 be a hurwitz system with values in @xmath51 .
suppose that @xmath45 .
let @xmath421 be the subgroup of @xmath51 generated by @xmath422 .
then for every @xmath423 the given hurwitz system is braid - equivalent to @xmath424 let us fix @xmath425 .
let @xmath426 be the subset consisting of elements @xmath48 such that the statement of the lemma holds for an arbitrary pair @xmath427 .
_ we claim @xmath428 is a subgroup of @xmath421_. let @xmath429 and let @xmath430
. then @xmath431 , so by assumption @xmath432 can be obtained from @xmath433 by a sequence of braid moves .
then one can obtain @xmath433 from @xmath434 by the inverse sequence of braid moves . thus @xmath435 .
if @xmath436 , then @xmath437 , so @xmath438 .
_ for every @xmath439 the element @xmath440 belongs to @xmath428_. applying a sequence of braid moves @xmath441 we can move the adjacent pair @xmath46 wherever we want among the first @xmath5 elements of the hurwitz system without changing the other elements .
so move @xmath46 to the left side of @xmath440 .
then we have @xmath442 .
we then move the pair @xmath443 with @xmath444 back to the initial position .
[ stepa3 ] _ for
every @xmath376 the element @xmath445 belongs to @xmath428_. first suppose that @xmath446 . in this case
let us perform a braid move @xmath448 .
one obtains @xmath449 where @xmath450 with @xmath451 and @xmath452 with @xmath453 .
we have @xmath454 , so @xmath455 .
let us move @xmath456 to the first place using @xmath457 .
one obtains @xmath458 . here
@xmath459 where @xmath445 .
let us perform again @xmath448 .
the elements @xmath460 have not changed with respect to the original hurwitz system , so @xmath461\mu_{k } ' = \mu_{k}$ ] . moving the second element @xmath462 to the first place by @xmath457 we obtain @xmath463 .
since @xmath464 we must have @xmath465 .
this proves @xmath466 with @xmath445 .
one extends this braid equivalence to every adjacent pair @xmath46 with @xmath45 by moving first the pair to the front , applying the braid equivalence we have just proved and moving the obtained pair back to the original position .
_ for every @xmath376 the element @xmath467 belongs to @xmath428_. the proof is the same as that of step [ stepa3 ] .
one uses the braid move @xmath468 instead of @xmath448 . by the preceding steps it remains to verify that @xmath469 for @xmath470 belong to the subgroup @xmath471 generated by @xmath472 where @xmath473 and @xmath376 .
we prove this by induction on @xmath474 .
we have @xmath475 , so @xmath476 . furthermore @xmath477\mu_{1}=\lambda_{1}\mu_{1 } \lambda_{1}^{-1}$ ] .
so @xmath478 and @xmath479 .
let @xmath480 .
suppose , by inductive assumption , that @xmath481 belong to @xmath482 .
then @xmath483\in h_2 $ ] .
we have @xmath484 , @xmath485 , so @xmath486 and therefore @xmath487 .
furthermore @xmath488\mu_{k}u_{k-1}^{-1 } $ ] .
therefore @xmath489 , so @xmath490 and @xmath491 .
the lemma is proved .
so far we have not made any restrictions on the type of the covering @xmath134 .
we shall be further occupied mainly with simply branched coverings with connected @xmath54 .
these coverings correspond to hurwitz systems @xmath33 with local monodromies @xmath492 , @xmath493 equal to transpositions and with transitive monodromy group @xmath494 .
we call such coverings _ simple coverings_. by hurwitz formula @xmath495 for a simple covering .
[ s4.37 ] let @xmath3 be a smooth projective curve .
we denote by @xmath496 and @xmath11 the hurwitz spaces which parametrize equivalence classes @xmath171 $ ] and @xmath28 $ ] respectively of simple coverings of degree @xmath2 branched in @xmath5 points ( cf .
[ s2.11 ] ) . as a first application of theorem [ s2.30c ] we give a proof , using braid moves , of the following well - known fact .
[ s4.37aa ] let @xmath497 .
the hurwitz spaces @xmath498 and @xmath499 which parametrize double coverings of @xmath3 branched in @xmath5 points are both irreducible . since @xmath498 and @xmath499 are smooth it suffices to prove they are connected .
let us first consider @xmath498 . by theorem [ s2.30c ] it suffices to prove that every hurwitz system @xmath33 with @xmath500 is braid - equivalent to @xmath501 . according to theorem [ s2.27 ] one has @xmath502 , @xmath503 and the same formulas hold for @xmath389 and @xmath391 .
hence whenever some @xmath504 or @xmath505 we can perform a braid move @xmath506 or @xmath448 respectively or both in order to obtain a new hurwitz system with @xmath507 .
the connectedness of @xmath498 implies the connectedness of the zariski dense open subset @xmath508 .
therefore @xmath499 is connected as well .
given an ordered @xmath5-tuple of permutations @xmath509 whose product @xmath510 , performing an elementary move @xmath379 or @xmath387 one obtains a new ordered @xmath5-tuple @xmath511 which has the same product @xmath512 .
we shall also call ordered @xmath5-tuples sequences .
[ s4.37ab ] two ordered @xmath5-tuples ( or sequences ) of permutations @xmath513 and @xmath514 are called braid - equivalent if @xmath515 is obtained from @xmath516 by a finite sequence of braid moves of type @xmath379 or @xmath387 .
we write @xmath517 .
notice the difference between braid equivalence of sequences and that of hurwitz systems ( cf . definition [ s3.31ab ] ) .
the latter includes also braid moves of the types @xmath518 .
[ s4.37b ] we recall some results proved in @xcite . given a permutation @xmath519
one considers an ordered @xmath5-tuple of transpositions @xmath513 such that @xmath520 .
let @xmath521 be the domains of transitivity of the group @xmath522
. then @xmath523 .
for reader s convenience we give a proof of the following important lemma due to mochizuki ( cf .
@xcite lemma 2.4 ) .
[ s4.37a ] let @xmath524 be as above .
. then @xmath526 and @xmath527 .
if @xmath46 is a pair such that @xmath528 with @xmath529 , then @xmath530 .
this shows that performing a sequence of braid moves ( of type @xmath441 ) one may replace @xmath516 by a concatenation of sequences @xmath531 where @xmath532 is formed by all transpositions of @xmath516 which belong to @xmath533 ordered in the way they appear in @xmath516 ( if @xmath534 we let @xmath535 ) .
furthermore we may move any @xmath532 to the first place or to the last place .
this shows that it suffices to prove the lemma in the particular case when @xmath536 is a transitive group .
let @xmath537 where @xmath538 .
one may vary the sequence within the set of sequences braid - equivalent to @xmath516 . given a sequence one may vary @xmath539
so that @xmath537 and finally given a sequence and a @xmath539 one may vary the factorization of @xmath539 . let @xmath515 , @xmath540 with the property @xmath541 and a factorization @xmath542 be chosen so that @xmath543 is minimal possible . if @xmath544 one has that @xmath545 is one of the transpositions of @xmath515 .
moving @xmath545 to the front by subsequent elementary moves @xmath387 one obtains a braid - equivalent sequence of the type @xmath546 . moving @xmath545 to the end by subsequent elementary moves @xmath379 one obtains @xmath547 as required in the lemma .
suppose that @xmath548 .
let @xmath549 .
the minimality of @xmath543 implies that @xmath550 for @xmath551 , so @xmath552 for @xmath553 are @xmath543 different transpositions of the sequence @xmath515 and every @xmath554 for @xmath555 enters in at most two transpositions of the set @xmath556 . applying several elementary moves one places @xmath557 adjacent to @xmath558 .
one obtains a sequence either of the type @xmath559 or of the type @xmath560 in which none of @xmath561 has been changed . in the first case one
has @xmath562 and in the second case @xmath563 .
in both cases one obtains a braid - equivalent sequence for which the sequence @xmath564 is replaced by @xmath565 .
this contradicts the minimality of @xmath543 .
[ s4.40a ] let @xmath566 be an @xmath5-tuple of transpositions and let @xmath520 .
let @xmath567 be the domains of transitivity of the cyclic group @xmath568 where @xmath569 .
following @xcite pp.369,370 if @xmath570 is the corresponding product of independent cycles we may write @xmath571 if @xmath572 and @xmath573 if @xmath574 .
such a representation is uniquely determined by @xmath575 if we assume that that @xmath576 is the minimal number among @xmath577 for each @xmath91 and if we order @xmath577 in such a way that @xmath578 .
let @xmath572 and let @xmath579 be the sequence @xmath580 . if @xmath574 one lets @xmath581 .
let @xmath582 be the concatenation of sequences @xmath583 .
the sequence @xmath582 consists of @xmath584 transpositions whose product equals @xmath575 .
the following theorem is proved in @xcite pp.369,370 .
it can also be deduced from the earlier paper of kluitmann @xcite .
[ s4.40 ] given a sequence of transpositions @xmath513 satisfying @xmath520 and such that @xmath585 is a transitive group ( therefore = @xmath6 ) , there is a braid - equivalent sequence @xmath586 which has the following form @xmath587 where @xmath588 and 1 . if @xmath589 , then @xmath590 for @xmath591 2 . if @xmath592 then @xmath593 where each @xmath594 appears two times if @xmath595 and @xmath596 appears an even number of times . [ s4.41 ] let @xmath513 be a sequence of transpositions in @xmath6 which generate a transitive group .
let @xmath597 and let @xmath567 be the domains of transitivity of the cyclic group @xmath568 .
suppose that @xmath598 .
let @xmath599 where @xmath551 .
then @xmath516 is braid - equivalent to a sequence @xmath600 .
let us consider the group @xmath601 which permutes the indices of @xmath602 .
let @xmath603 be transpositions in @xmath601 which generate it .
we let @xmath604 and choose arbitrarily @xmath605 for each @xmath606 .
if @xmath607 we consider the pair of transpositions @xmath608 .
we then consider a sequence of @xmath5 transpositions in @xmath6 which is a concatenation of the sequence @xmath582 , of the pairs associated with @xmath609 and of the sequence @xmath610 with @xmath611 repeated an even number of times .
we obtain a sequence @xmath612 with product @xmath575 which generates a transitive group . according to theorem [ s4.40 ]
this sequence is braid - equivalent to @xmath516 . moving the pair @xmath613 to the end we obtain the required sequence .
[ s4.41a ] given a permutation @xmath86 whose domains of transitivity are @xmath567 with @xmath614 we let @xmath615 .
let us consider a sequence of transpositions @xmath616 .
let @xmath617 be the domains of transitivity of the group @xmath618 .
we let @xmath619 .
[ s4.42 ] let @xmath513 be a sequence of transpositions and let @xmath520
. then @xmath620 and @xmath621 .
if @xmath622 then there is a sequence @xmath515 braid - equivalent to @xmath516 such that @xmath623 and @xmath624 generate the same group as @xmath625 .
it is clear that every @xmath577 belongs to some @xmath626 .
furthermore one may replace @xmath516 by a braid - equivalent sequence which is a concatenation @xmath627 of sequences @xmath628 whose elements belong to @xmath533 ( cf .
the proof of lemma [ s4.37a ] ; if @xmath629 one lets @xmath535 ) .
it thus suffices to prove the lemma in the case when @xmath536 is transitive , which we further assume , so @xmath630 .
let us consider the braid - equivalent sequence @xmath515 of theorem [ s4.40 ] .
the subsequence @xmath582 has @xmath631 elements .
the remaining @xmath632 elements appear in pairs and their number is at least @xmath633 since @xmath634 is a transitive group .
if @xmath635 the pair @xmath636 is repeated at least twice , so canceling it from the sequence does not change the group generated by the transpositions @xmath637 .
one has @xmath638 , so @xmath639 . adding @xmath640 to both sides of the inequality @xmath641 we obtain the equivalent inequality @xmath642 . [ s4.43 ] a transitive subgroup @xmath643 is called _ imprimitive _ if there is a decomposition @xmath644 such that for @xmath645 and @xmath139 one has @xmath646 for some @xmath647 and furthermore @xmath648 for @xmath139 where @xmath649 . a transitive subgroup @xmath142 which is not imprimitive
is called _ primitive_. if @xmath2 is a prime number clearly every transitive subgroup of @xmath6 is primitive
. [ s4.43a ] suppose that @xmath142 is a primitive group which contains a transposition
. then @xmath650 . in particular
if @xmath2 is a prime number every transitive subgroup of @xmath6 which contains a transposition is equal to @xmath6 .
let us prove that if @xmath51 contains a nontrivial symmetric subgroup @xmath651 with @xmath652 then @xmath51 contains a @xmath653 with @xmath654 .
indeed , since @xmath51 is primitive there are @xmath655 such that @xmath656 . hence if @xmath657 one has @xmath658 .
let @xmath659 .
then @xmath660 .
so @xmath661 and @xmath662 is contained in @xmath51 .
we conclude the proof of the lemma by induction on @xmath663 starting from @xmath664 with @xmath665 . to each connected component of @xmath496
is associated a fixed monodromy group @xmath142 and similarly to each connected component of @xmath11 is associated a conjugacy class of subgroups .
[ s4.44 ] let @xmath666 .
we denote by @xmath667 and @xmath668 the union of connected components of @xmath669 and @xmath670 respectively which correspond to coverings with primitive monodromy groups ( therefore equal to @xmath6 by lemma [ s4.43a ] ) .
the following theorem is due in the case @xmath0 to graber , harris and starr @xcite .
[ s4.45 ] let @xmath3 be a smooth , projective curve .
let @xmath671 .
if @xmath21 the hurwitz spaces @xmath667 and @xmath1 are irreducible .
the case @xmath672 is classical and due to clebsch and hurwitz @xcite , @xcite .
modern proofs may be found in @xcite p.368 and @xcite p.197 .
suppose that @xmath111 .
let us first consider @xmath667 . since @xmath667 is smooth in order to prove its irreducibility it suffices to prove it is connected . by theorem [ s2.30c ]
it suffices to show that every hurwitz system @xmath33 with monodromy group @xmath6 is braid - equivalent to @xmath673 where each @xmath674 appears twice if @xmath675 and @xmath676 appears @xmath677 times .
[ b1 ] we claim that every hurwitz system @xmath33 is braid - equivalent to a hurwitz system @xmath678 such that @xmath679 is a transitive group .
indeed , let @xmath680\cdots [ \lambda_{g},\mu_{g}]$ ] and let @xmath617 be the domains of transitivity of the group @xmath618 .
one has @xmath681 , so if @xmath682 is not transitive then @xmath683 . in this case
the inequality @xmath684 is satisfied , so according to lemma [ s4.42 ] one may replace @xmath566 by a braid - equivalent sequence @xmath586 such that @xmath685 and @xmath686 , so @xmath687 .
suppose that @xmath688 .
let @xmath423 be such that @xmath689 . then according to main lemma [ s3.32 ] the hurwitz system @xmath690 is braid - equivalent to @xmath691 . if @xmath692 are the domains of transitivity of @xmath693 then clearly @xmath694 . repeating this argument after a finite number of steps one obtains a braid - equivalent hurwitz system @xmath678 with transitive @xmath679 .
[ b2 ] let @xmath33 be a hurwitz system .
suppose that @xmath695 .
we claim there is a braid - equivalent sequence @xmath696 such that @xmath697 .
let us first suppose that @xmath698 .
decomposing @xmath699 into a product of nontrivial independent cycles and choosing one of them @xmath700 we have @xmath701 .
according to step [ b1 ] we may replace @xmath33 by a braid - equivalent system @xmath678 such that @xmath702 generate a transitive group . then according to mochizuki s lemma [ s4.37a ] we may replace @xmath703 by a braid - equivalent sequence @xmath704 . applying the braid move @xmath705 ( cf .
we transform @xmath699 into @xmath706 .
suppose that @xmath707 .
we reason in the same way and use the braid move @xmath708 which transforms @xmath709 into @xmath710 according to corollary [ s2.30a ] ( i ) . similarly if @xmath711 and @xmath712 for @xmath713 we have @xmath714 and applying the braid move @xmath468 we transform @xmath715 into @xmath716 thus decreasing @xmath717 . if @xmath505 , @xmath712 for @xmath713 and @xmath718 we apply @xmath448 to transform @xmath719 into @xmath720 thus decreasing @xmath721 . starting with @xmath33 and applying
several times step [ b2 ] one obtains a braid - equivalent hurwitz system @xmath722 . here
@xmath723 , so applying the argument in the case @xmath724 ( see e.g. @xcite p.368 , or @xcite p.197 ) one obtains the initial hurwitz system is braid - equivalent to eq .. the connectedness of both @xmath667 and @xmath1 follows now from theorem [ s2.30c ]
. theorem [ s4.45 ] may be generalized in a straightforward manner to coverings @xmath134 which have simple ramifications in all except possibly one discriminant point .
the conjugacy classes of @xmath6 correspond bijectively to partitions of @xmath2 , namely @xmath16 where @xmath17 and @xmath18 . to each partition one
associates the orbit of the permutation @xmath725 given a partition @xmath726 let us denote by @xmath727 and @xmath728 ) the hurwitz spaces which parametrize equivalence classes @xmath171 $ ] and @xmath729 $ ] respectively , where @xmath54 is connected and @xmath134 is a covering of degree @xmath2 with @xmath5 discriminant points of simple ramification and one additional discriminant point whose local monodromy belongs to the conjugacy class of eq .. let @xmath730 and @xmath19 parametrize such coverings with primitive monodromy group .
[ s4.49 ] let @xmath3 be a smooth , projective curve .
let @xmath671 .
suppose that @xmath21 .
let @xmath726 be an arbitrary partition of @xmath2 .
then the hurwitz spaces @xmath730 and @xmath19 are irreducible .
if @xmath726 is the trivial partition ( that is @xmath574 for @xmath139 ) this is the content of theorem [ s4.45 ] .
suppose that @xmath731 .
let us denote the permutation eq .
by @xmath732 . the theorem will be proved if we can show that every hurwitz system @xmath733 of the considered type may be reduced after a finite number of braid moves of the types @xmath734 to the normal form @xmath735 where the @xmath5-tuple of transpositions @xmath736 is the one defined in [ s4.40a ] and theorem [ s4.40 ] .
first using braid moves of type @xmath379 we may replace the original hurwitz system by one for which @xmath737 belongs to the orbit of @xmath738 .
we then let @xmath680\cdots [ \lambda_{g},\mu_{g}]t^{-1}_{n+1}$ ] and repeating the arguments of the proof of theorem [ s4.45 ] we obtain a braid - equivalent hurwitz system @xmath739 with @xmath740 .
we are further allowed to apply only braid moves of types @xmath379 and @xmath387 .
the next step is to replace the obtained hurwitz system by one in which at the @xmath741-th place stays @xmath742 .
in fact the permutation @xmath742 has the same cyclic type as @xmath743 , so @xmath744 .
let @xmath745 where @xmath746 are transpositions . by the hypothesis and lemma [ s4.43a ] we have @xmath747 .
so using mochizuki s lemma [ s4.37a ] we may replace @xmath703 by a braid - equivalent @xmath5-tuple @xmath748 .
we then have @xmath749 , so @xmath750 . repeating this argument with @xmath751
we obtain a braid - equivalent @xmath741-tuple whose @xmath741-th element is @xmath742 and the product of the first @xmath5 equals @xmath738 . using a finite number of braid moves among the first @xmath5 transpositions we obtain the required normal form according to theorem [ s4.40 ] .
so far we worked with coverings with primitive monodromy groups .
we now wish to consider the imprimitive case .
[ s4.52 ] let @xmath643 be a transitive imprimitive subgroup which contains a transposition .
then there is a unique decomposition @xmath752 as in definition [ s4.43 ] so that @xmath753 is a primitive subgroup of @xmath754 for @xmath139 .
furthermore the following properties hold . 1 .
all transpositions contained in @xmath51 form one conjugacy @xmath51-orbit @xmath755 .
2 . @xmath756 .
3 . @xmath757 for @xmath139 .
let @xmath665 and let @xmath758 .
let @xmath759 and let @xmath760 be the domains of transitivity of @xmath421 . since @xmath421 is a normal subgroup every @xmath63 permutes the orbits @xmath761 .
furthermore @xmath762 , so @xmath763 .
all transpositions which belong to @xmath421 form one @xmath51-orbit since this property holds for @xmath754 .
let @xmath764 be an arbitrary transposition of @xmath51 .
we claim it is impossible that @xmath765 for @xmath551 . indeed , if this were the case letting @xmath766 we would have @xmath767 , so @xmath646 . if @xmath768 , then @xmath769 and on the other hand @xmath770 , so @xmath771 which is absurd .
we obtain @xmath772 for some @xmath91 , so @xmath773 .
this proves ( i ) .
suppose now that @xmath774 is an arbitrary decomposition as required in definition [ s4.43 ] .
let @xmath775 be a transposition in @xmath51 .
then the argument above shows that @xmath776 for some @xmath91 .
this shows that @xmath777 where @xmath778 .
thus each orbit @xmath779 of @xmath759 is contained in some orbit @xmath780 .
assuming @xmath781 is primitive subgroup of @xmath782 we conclude that @xmath783 .
this proves the uniqueness statement .
the other properties were already proved .
[ s4.54 ] let @xmath134 be a covering of smooth , irreducible , projective curves of degree @xmath2 .
suppose that at least one of the discriminant points is simple and suppose that the monodromy group of the covering is @xmath784 .
then there exists a unique decomposition @xmath785 , where @xmath786 for @xmath787 and @xmath788 has primitive monodromy group .
if @xmath53 has @xmath5 simple discriminant points so does @xmath788 and @xmath789 induces a bijective correspondence between these two sets .
one chooses an unramified point @xmath671 , and a bijection @xmath790 .
one applies lemma [ s4.52 ] to the monodromy group of @xmath791 and obtains a decomposition @xmath792 . replacing @xmath96 by another bijection results in replacing the monodromy group @xmath51 by a conjugate one @xmath793 .
the conjugation by @xmath86 transforms the @xmath51-orbit of transpositions contained in @xmath51 to the @xmath794-orbit of transpositions contained in @xmath794 .
therefore @xmath575 transforms the corresponding decompositions of @xmath89 : @xmath795 .
this proves the uniqueness of the decomposition @xmath792 with the required properties .
the last statement of the corollary is obvious .
[ s4.55 ] let @xmath3 be a smooth , projective curve of genus @xmath4 .
. if @xmath2 is prime then @xmath797 .
if @xmath2 is not prime let @xmath798 be its maximal divisor @xmath799 .
suppose that @xmath800 .
then the connected components of @xmath11 which correspond to simple coverings with monodromy groups @xmath10 are in one - to - one correspondence with the equivalence classes of unramified coverings @xmath25 $ ] of degrees @xmath801 where @xmath802 .
if @xmath2 is prime every transitive subgroup of @xmath6 is primitive , so @xmath797 .
suppose that @xmath2 is not prime .
let @xmath134 be a simple covering of degree @xmath2 , simply ramified in @xmath5 points , with monodromy group @xmath10 .
then the decomposition @xmath803 of corollary [ s4.54 ] is with tale @xmath789 .
furthermore by the uniqueness of this decomposition if @xmath804 defines an equivalence of coverings of @xmath3 then @xmath805 is induced by an equivalence @xmath806 . given an tale covering @xmath807 of degree @xmath808 we consider the hurwitz space @xmath809 where @xmath810 .
it is connected by theorem [ s4.45 ] .
we then consider the zariski open subset consisting of @xmath811 $ ] with discriminant points belonging to different fibers of @xmath812 .
every such @xmath813 yields a simple covering of @xmath3 . in this way one obtains a connected component of @xmath11 whose conjugacy class of monodromy groups is imprimitive .
vice versa by corollary [ s4.54 ] every connected component of @xmath11 whose associated conjugacy class of monodromy groups is imprimitive is obtained in this way . [ s4.57 ] let @xmath814 and let @xmath3 be a curve of genus @xmath815 .
the connected unramified coverings of @xmath3 of degree 2 are classified up to equivalence by the points of order 2 in the jacobian @xmath816 . applying theorem [ s4.55 ]
we obtain if @xmath231 there are @xmath817 different connected components of @xmath818 which parametrize coverings with imprimitive monodromy group .
the monodromy group in this case is isomorphic to the dihedral group @xmath819
unless otherwise specified in this section we shall work with hurwitz systems @xmath820 where @xmath821 and @xmath625 are transpositions in @xmath6 .
let @xmath822 be a sequence of transpositions in @xmath6 and let @xmath597 .
suppose that the subgroup @xmath823 has only one nontrivial domain of transitivity @xmath64 and let @xmath824 .
we have the inequality @xmath825 ( cf .
lemma [ s4.42 ] ) so if @xmath824 is fixed the minimum for @xmath5 is reached for @xmath826 , and then @xmath827 .
this happens if and only if @xmath597 is a cycle of order @xmath828 .
[ s5.58bis ] we call a sequence of transpositions @xmath822 _ minimal _ if @xmath597 is a cycle of order @xmath828 .
[ s5.59bis ] let @xmath822 be a sequence of transpositions of @xmath6 such that @xmath823 has a unique nontrivial domain of transitivity @xmath64 .
suppose that @xmath755 is not minimal .
then for any @xmath829 there exists a braid - equivalent sequence @xmath830 such that @xmath685 for some @xmath831 . without loss of generality
we may assume that @xmath832 .
let @xmath597 and let @xmath567 be the domains of transitivity of @xmath575 . replacing @xmath566 by a braid - equivalent sequence we may assume it has the normal form of theorem [ s4.40 ] .
the non - minimality hypothesis @xmath833 implies that @xmath755 contains @xmath834 pairs of the form @xmath835 if @xmath589 or @xmath594 with @xmath836 if @xmath598 .
consider the cyclic group @xmath568 .
if @xmath837 then @xmath838 , so the sequence @xmath839 is braid - equivalent to @xmath566 ( cf .
@xcite , @xcite ) .
if @xmath589 we may find @xmath837 such that @xmath840 .
here we let @xmath841 . if @xmath598 and @xmath842 we find @xmath843 such that @xmath840 and we let @xmath844 .
if @xmath845 with @xmath237 we find @xmath837 such that @xmath846 and we let @xmath847 . in each case
the sequence @xmath839 contains the adjacent pair @xmath613 .
moving it to the end we obtain the sequence required in the lemma .
[ s5.60bis ] given a hurwitz system @xmath33 where @xmath413 let @xmath848 be the domains of transitivity of @xmath849 ordered in such a way that @xmath850 .
let @xmath851 be the associated partition of @xmath2 .
given @xmath779 let @xmath532 be the subsequence of @xmath822 composed of those transpositions which move points of @xmath779 .
let @xmath852 .
we call @xmath779 minimal or non - minimal if the sequence @xmath532 is minimal or non - minimal respectively
. the trivial case @xmath853 is assumed minimal .
[ s5.61bis ] let @xmath33 be a hurwitz system with monodromy group @xmath6 .
let @xmath680\cdots [ \lambda_{g},\mu_{g}]$ ] and let @xmath854 .
let @xmath617 be the domains of transitivity of @xmath855 ordered as in [ s5.60bis ] .
suppose that @xmath856 is maximal in lexicographic order among the partitions associated with all hurwitz systems of the type @xmath857 braid - equivalent to the given one .
then @xmath858 ( equivalently @xmath682 is transitive ) if and only if @xmath859 . if @xmath860 then @xmath861 . furthermore @xmath862 and for each @xmath237 the @xmath855-orbit @xmath779 is minimal , equivalently @xmath863 .
in particular if @xmath864 then @xmath618 has a single nontrivial domain of transitivity .
we proved in lemma [ s4.42 ] that if @xmath682 is transitive then @xmath859 . that the maximality of @xmath856 and @xmath859 imply that @xmath858 is a fact evident from the proof of theorem [ s4.45 ] ( cf .
step [ b1 ] ) .
we shall further assume that @xmath860 , so @xmath855 is not transitive .
suppose , by way of contradiction , that there is a non - minimal @xmath865 with @xmath480 .
let @xmath866 , so @xmath867 .
let @xmath868 , so @xmath869 .
for every @xmath870 one has @xmath871 .
let @xmath872 and @xmath873 be chosen in such a way that @xmath874 and furthermore the length of the factorization @xmath875 with @xmath876 or @xmath877 is minimal possible among all @xmath870 and all @xmath63 satisfying @xmath878 .
we claim @xmath544 and @xmath879 .
in fact by minimality we have @xmath880 and for each @xmath881 the adjacent @xmath882 do not belong to the same subgroup @xmath428 or @xmath877 .
suppose that @xmath548 . then by minimality @xmath883 . if @xmath884 we might replace @xmath885 by @xmath886 , so @xmath887 .
the group @xmath428 acts trivially on @xmath888 , therefore since @xmath889 we have @xmath890 .
hence @xmath891 where @xmath892 has a shorter factorization then @xmath48 .
this is an absurd , so @xmath544 .
we have thus proved there exists an @xmath872 and @xmath893 such that @xmath894 for some @xmath895 . according to lemma [ s5.59bis ]
we may replace @xmath865 by a braid - equivalent sequence which contains the adjacent pair @xmath613 for some @xmath870 . by main lemma [ s3.32 ] replacing @xmath613 by @xmath896
one obtains a braid - equivalent hurwitz system .
the following cases may occur . _
@xmath897 for some @xmath898_.if @xmath261 then the new hurwitz system has @xmath899 as a domain of transitivity , so the corresponding partition eq . is greater than @xmath856 in lexicographical order . if @xmath900 we interchange @xmath885 and @xmath901 and make the same conclusion . in both cases
we obtain a contradiction with the maximality of @xmath856 .
@xmath902.let @xmath903 .
moving the pair @xmath904 to the right of @xmath532 we obtain a braid - equivalent sequence @xmath905 with the same domain of transitivity as @xmath532 .
canceling the pair from the obtained hurwitz system does not change the monodromy group @xmath51 .
let us choose @xmath906 such that @xmath907 . applying main lemma [ s3.32 ] we obtain a new hurwitz system @xmath678 whose associated partition @xmath856 has the property that its first @xmath91 terms are greater in lexicographical order then the first @xmath91 terms of @xmath856 .
this contradicts the maximality of the partition @xmath856 .
the claim that @xmath532 is minimal for every @xmath237 is proved .
according to lemma [ s4.42 ] we have @xmath908 and if this inequality is strict we can apply the argument of step [ b1 ] of theorem [ s4.45 ] in order to increase @xmath856 which is impossible .
thus @xmath909 . finally if @xmath864 then @xmath910 , so minimality of @xmath779 means @xmath911 .
therefore the maximality of @xmath856 implies that only @xmath912 may be nontrivial domain of transitivity .
[ s5.63bbis ] given a hurwitz system
@xmath33 we let @xmath913 , @xmath914 , @xmath915 , @xmath916 ( cf .
[ s4.41a ] ) .
we consider the class @xmath917 of hurwitz systems braid - equivalent to the given one . when dealing with the problem of finding a system in @xmath917 of a simplest form it makes sense to assume ( without loss of generality ) that @xmath33 satisfies the following property : ( * ) _ @xmath918 is minimal in lexicographic order among the hurwitz systems in @xmath917 and furthermore @xmath856 ( cf .
[ s5.60bis ] ) is maximal in lexicographic order among the hurwitz systems in @xmath917 of the type @xmath857 _ [ s5.64bis ] let @xmath919 .
let @xmath3 be a smooth , projective curve of genus @xmath815 .
then the hurwitz space @xmath1 parametrizing simple coverings with monodromy groups equal to @xmath6 is irreducible . by theorem [ s2.30c ]
it suffices to show that the equivalence class ( modulo inner automorphisms ) of every hurwitz system @xmath33 with monodromy group @xmath6 may be reduced by a finite number of braid moves to the normal form @xmath920\ ] ] where each @xmath674 appears twice for @xmath921 . without loss of generality
we may assume that @xmath33 has minimal @xmath918 and maximal @xmath856 as in [ s5.63bbis ] ( * ) [ c1 ] _ _ we claim @xmath922__let us first suppose that @xmath698 .
let @xmath545 be a transposition such that @xmath701 .
let @xmath680\cdots [ \lambda_{g},\mu_{g}]$ ] .
first suppose that @xmath923 .
since @xmath924 we have @xmath925 , thus @xmath859 . by lemma [ s5.61bis ]
@xmath618 is a transitive group , so by mochizuki s lemma [ s4.37a ] we can replace @xmath566 by a braid - equivalent sequence @xmath586 such that @xmath926 . applying @xmath927 ( cf .
one transforms @xmath699 into @xmath706 .
this contradicts the minimality of @xmath918 .
suppose that @xmath864 . then according to lemma [ s5.61bis ] the subgroup @xmath585 has two orbits : @xmath928 where @xmath929 .
moreover replacing if necessary @xmath566 by a braid - equivalent sequence we may assume that @xmath930 for every @xmath931 .
varying over all hurwitz systems with @xmath932 fixed and braid - equivalent to the given one we want to figure out which transpositions may appear at the first place .
let @xmath933 . if @xmath934 then by mochizuki s lemma [ s4.37a ] one may replace the sequence @xmath566 by a braid - equivalent one @xmath935 .
let @xmath936 , @xmath937 . by hypothesis @xmath938 so @xmath939 such that @xmath940 .
let @xmath941 , so @xmath942 with @xmath943 .
by main lemma [ s3.32 ] the hurwitz system @xmath944 is braid - equivalent to the given one . applying again mochizuki s lemma [ s4.37a ]
we see that every @xmath945 with @xmath946 may be placed first in some braid - equivalent hurwitz system .
we may vary @xmath947 with the property @xmath940 , or equivalently with the property @xmath948 .
if the orbit @xmath949 has @xmath950 elements , then an arbitrary transposition @xmath539 may be placed first in some braid - equivalent hurwitz system @xmath857 .
doing this for @xmath951 and applying the braid move @xmath927 we transform @xmath699 into @xmath706 thus obtaining a contradiction with the minimality of @xmath918 .
it remains to consider the case @xmath952 . in this case @xmath953 .
if @xmath954 or @xmath955 with @xmath956 , @xmath957 we might decrease @xmath958 transforming @xmath699 into @xmath706 with an appropriate @xmath545 as we saw above and this contradicts the minimality of @xmath918 .
it remains to consider the case @xmath959 .
let @xmath960 . replacing the @xmath5-tuple @xmath566 by a braid - equivalent one
we may assume that it equals @xmath961 .
since @xmath962 we have @xmath963=1 $ ] .
so @xmath964 .
applying this braid move we obtain @xmath965 .
we may replace @xmath966 by the braid - equivalent pair @xmath967 and then apply @xmath705 transforming @xmath699 into @xmath968 .
this contradicts again the minimality of @xmath918 .
this proves @xmath969 . if @xmath970 , but @xmath504 we have @xmath971\cdots [ \lambda_{k-1},\mu_{k-1}]=1 $ ] .
the braid move @xmath468 transforms @xmath972 into @xmath973 according to theorem [ s2.27 ] so the same argument as above may be applied proving that @xmath504 contradicts the minimality of @xmath918 .
therefore @xmath974 .
_ we claim the group @xmath849 has orbits @xmath975 , @xmath976 and there exists @xmath977 $ ] such that every @xmath978 equals either 1 or @xmath979_. for every hurwitz system with @xmath980 we have by theorem [ s2.27 ] ( b ) that @xmath448 transforms @xmath981 into @xmath982 .
assume that @xmath983 and @xmath505 .
upon a substitution @xmath984 we may use arguments similar to those of step [ c1 ] trying to decrease @xmath721 and thus obtaining a contradiction with the minimality of @xmath918 .
we are not allowed however to use braid moves of the types @xmath384 and @xmath391 since these change @xmath972 while we want to preserve at every braid move the equality @xmath974 .
we thus have the following : @xmath849 has orbits @xmath975 , @xmath976 , @xmath985 , @xmath986 . here if @xmath987 we have either @xmath988 or @xmath989 with @xmath990 .
if @xmath991 or @xmath992 respectively we might apply mochizuki s lemma [ s4.37a ] and the braid move @xmath993 in order to decrease @xmath994 .
this is impossible by the minimality of @xmath918 so for every @xmath995 either @xmath996 or @xmath997 [ c3 ] _ we claim @xmath996 for @xmath998 , @xmath999 _ .
since renumbering and braid moves are commutative operations we may assume without loss of generality that @xmath1000 and @xmath1001 so every @xmath978 equals either 1 or @xmath676 .
let @xmath474 be the minimal index such that @xmath1002 .
if @xmath1003 there is nothing to prove , so suppose that @xmath1004 .
replacing eventually @xmath566 by a braid - equivalent sequence we may assume that @xmath33 has the following form : @xmath1005 first suppose that @xmath1006 .
we claim that replacing @xmath1007 by @xmath676 one obtains a braid - equivalent hurwitz system .
indeed @xmath506 ( cf .
corollary [ s2.30a ] ) transforms eq . into @xmath1008 .
apply the following sequence of braid - equivalences : replace @xmath1009 by @xmath1010 ; move this pair to the end by braid moves of type @xmath379 ; perform @xmath1011 ; move the pair @xmath1010 backward to the front by braid moves of type @xmath387 ; replace it by @xmath1009 and finally apply @xmath468 .
one obtains @xmath1012 as claimed .
we may thus suppose that eq .
equals eq .. applying @xmath1013 to eq . and
replacing @xmath1009 by @xmath1014 we obtain @xmath1015 applying @xmath1016 one transforms @xmath1017 into 1 without changing @xmath1018 . replacing @xmath1014 by @xmath1009 and applying @xmath1019
we obtain a hurwitz system with @xmath1020 , @xmath1021 and @xmath1022 or @xmath676 for @xmath995 .
this shows that the assumption @xmath505 for @xmath1004 contradicts the minimality of @xmath918 .
we have so far worked with hurwitz systems . when working with equivalence classes we may moreover renumber arbitrarily @xmath1023 .
we thus conclude that the equivalence class ( modulo inner automorphisms ) of a hurwitz system of the type of step [ c3 ] is braid - equivalent to the equivalence class eq . with @xmath1003 , so it has the normal form eq . as claimed
unless otherwise specified in this section we shall work with hurwitz systems @xmath1024 where @xmath231 and @xmath1025 are transpositions in @xmath6 .
[ s6.72 ] let @xmath1024 , with @xmath1026 , be a hurwitz system with monodromy group @xmath6 .
suppose that @xmath1027 .
let @xmath1028 .
then the equivalence class @xmath1029 $ ] is braid - equivalent to @xmath1030\ ] ] we may assume without loss of generality that @xmath1031 is minimal and @xmath856 is maximal in lexicographic order as in [ s5.63bbis ] ( * ) . by lemma [ s5.61bis ] the subgroup @xmath1032 has a single nontrivial domain of transitivity @xmath912 and furthermore since @xmath1033 we may assume that @xmath1034 for @xmath1035 . let @xmath1036 .
decomposing @xmath1037 into a product of independent cycles @xmath1038 we have each @xmath1039 contains at most one element of @xmath912 . indeed , otherwise using mochizuki s lemma [ s4.37a ] and applying the braid move @xmath1011 ( cf .
we might decrease @xmath1040 .
furthermore the transitivity of @xmath1041 implies that every @xmath1039 contains exactly one element of @xmath912 and each element of @xmath1042 is contained in one of @xmath1043 .
the inequality @xmath1044 , equivalent to @xmath1045 , implies that at least one element of @xmath912 does not appear in the cycles @xmath1046 .
we claim that if @xmath1047 then there is a braid - equivalent hurwitz system @xmath1048 such that @xmath1049 is a cycle of order @xmath1050 containing a single element of @xmath912 .
braid moves and renumbering of @xmath89 are commutative operations , so for proving the claim we may assume without loss of generality that @xmath1051 equals @xmath1052 performing @xmath1053 we obtain @xmath1054 moving the pair @xmath1055 to the end , replacing it by @xmath1056 and performing @xmath1057 we obtain @xmath1058 let us consider the system @xmath1059 moving the pair @xmath1060 to the front and then performing @xmath1053 we obtain @xmath1061 moving @xmath1062 to the end , replacing it by @xmath1063 and performing @xmath1057 we obtain @xmath1064 since eq . is braid - equivalent to eq .
and eq . is braid - equivalent to eq .
we conclude that eq .
is braid - equivalent to eq .. the sequence eq .
is obtained from eq . by moving one element from the cycle
@xmath1065 to the cycle @xmath1066 . repeating this transformation several times we obtain a cycle @xmath1049 of order @xmath1050 .
we thus proved that replacing the given hurwitz system by a braid - equivalent one we may assume that @xmath618 has a single nontrivial orbit @xmath1067 and @xmath1068 where @xmath1069 and @xmath1070 .
renumbering and normalizing the sequence @xmath566 according to the classical result of clebsch and hurwitz we obtain that @xmath1071 $ ] is braid - equivalent to the equivalence class eq .. [ s6.76a ] the proof of proposition [ s6.72 ] may be easily adapted in order to show that every equivalence class @xmath1072 $ ] with full monodromy group @xmath6 and @xmath1073 is braid - equivalent to @xmath1074.\ ] ] here one should modify condition [ s5.63bbis ] ( * ) assuming without loss of generality @xmath1075 is minimal in the reverse lexicographical order and @xmath856 is maximal in lexicographical order .
one then repeats the proof of the proposition using @xmath1076 instead of @xmath1053 and @xmath705 instead of @xmath1057 .
[ s6.76 ] let @xmath1024 be a hurwitz system with monodromy group @xmath6 .
let @xmath617 be the domains of transitivity of @xmath618 ordered in such a way that @xmath1077 .
suppose that @xmath1031 is minimal and @xmath856 is maximal as in [ s5.63bbis ] ( * ) .
suppose that @xmath1078 .
then the following three conditions can not hold simultaneously . 1 .
@xmath912 is non - minimal ( cf .
[ s5.60bis ] ) .
there exists @xmath1079 such that @xmath1080 ( equivalently there exists @xmath1081 such that @xmath1082 ) .
there exists @xmath1083 such that @xmath1084 .
decomposing @xmath1085 into a product of independent cycles @xmath1086 we see that each of the two conditions of ( ii ) is equivalent to the existence of a @xmath1087 which contains elements of both @xmath912 and @xmath1088 .
let @xmath1089 be the decomposition of @xmath1037 into a product of independent cycles .
the minimality of @xmath1031 implies that neither @xmath1087 nor @xmath1090 may contain two different elements of some @xmath1091 .
in fact if this were the case then we might apply mochizuki s lemma [ s4.37a ] and one of the braid moves @xmath705 or @xmath1057 in order to decrease @xmath1031 .
in particular for each @xmath1092 we have an alternative : either @xmath1093 or @xmath1094 .
suppose , by way of contradiction , that ( i ) , ( ii ) and ( iii ) hold simultaneously .
let @xmath1095 $ ] .
by lemma [ s5.61bis ] we must have @xmath860 since otherwise @xmath618 is transitive , so applying mochizuki s lemma [ s4.37a ] and the braid move @xmath705 we might decrease @xmath1096 which contradicts the minimality of @xmath1031 assumed in the lemma .
lemma [ s5.61bis ] yields moreover that @xmath862 , so condition
( i ) means @xmath1097 has @xmath598 domains of transitivity @xmath567 .
we claim one can choose @xmath885 and @xmath901 satisfying ( ii ) and ( iii ) respectively so that they belong to different domains .
in fact if @xmath1098 and all @xmath1092 with @xmath1094 belong to @xmath1099 then we may replace @xmath901 by an arbitrary element of some other @xmath1100 according to the alternative of the preceding paragraph .
let @xmath1101 .
let @xmath1102 be the subsequence of @xmath822 consisting of the transpositions that move elements of @xmath912 .
let @xmath1103 .
using corollary [ s4.41 ] we can replace @xmath1102 by a sequence @xmath1104 and then move the pair @xmath1105 to the end of @xmath755 .
we then perform the braid move @xmath1076 which transforms @xmath1106 into @xmath1107 .
the group generated by the new sequence @xmath1108 has orbit @xmath1109 . by mochizuki
s lemma [ s4.37a ] one may replace @xmath1110 by @xmath1111 . performing @xmath705
one transforms @xmath1085 into @xmath1112 and clearly @xmath1113 .
this contradicts the minimality of @xmath1031 .
[ s6.79 ] let @xmath1024 be a hurwitz system with monodromy group @xmath6 .
let @xmath1114 .
suppose that @xmath1031 is minimal and @xmath856 is maximal as in [ s5.63bbis ] ( * ) .
suppose that @xmath1095=1 $ ] .
then @xmath1026 and the equivalence class @xmath1029 $ ] is braid - equivalent to eq .. let @xmath617 be the orbits of @xmath618 .
let @xmath1028 . according to lemma [ s5.61bis ] we have @xmath1115 for each @xmath237 and @xmath1116 .
suppose , by way of contradiction , that @xmath1078 . according to lemma [ s6.76 ] its conditions ( i )
( iii ) can not hold simultaneously .
the orbit @xmath912 is non - minimal since @xmath1117 . as we saw in the course of the proof of lemma [ s6.76 ]
the minimality of @xmath1031 implies that if one decomposes @xmath1085 into a product of independent cycles @xmath1118 then every @xmath1087 contains at most one element of @xmath912 .
the inequality @xmath1044 implies that @xmath1119 .
thus there exists an element @xmath1083 such that @xmath1084
. therefore condition ( ii ) of lemma [ s6.76 ] must fail to hold , i.e. we have @xmath1120 for every @xmath1079 .
let @xmath1121 be the set of fixed points of @xmath1085 .
we have @xmath1122 so the group @xmath618 leaves @xmath408 invariant .
since @xmath1123=1 $ ] we have @xmath1124 .
therefore the assumption @xmath1078 implies that @xmath1125 is not a transitive group which is a contradiction . the last part of the lemma refers to proposition [ s6.72 ] .
[ s6.80 ] this lemma may be stated differently : if @xmath1123=1 $ ] and @xmath1078 then one can decrease @xmath1096 by a sequence of braid moves ( at the expense of possible increasing of @xmath1040 ) . modifying appropriately lemma [ s6.76
] one can prove similarly that given a hurwitz system @xmath1024 with full monodromy group @xmath6 and such that @xmath1114 , then @xmath1123=1 $ ] and @xmath1126 imply that one can decrease @xmath1040 by a sequence of braid moves ( at the expense of possible increasing of @xmath1096 ) .
by theorem [ s2.30c ] it suffices to show that given a hurwitz system @xmath1024 with monodromy group @xmath6 its equivalence class is braid - equivalent to @xmath1128.\ ] ] without loss of generality we may assume that @xmath1024 has minimal @xmath1031 and maximal @xmath856 as in [ s5.63bbis ] ( * ) . here
@xmath617 are the domains of transitivity of @xmath849 ordered in such a way that @xmath850 .
we aim to prove that @xmath1026 in order to apply proposition [ s6.72 ] .
let @xmath1095 $ ] .
our first goal is to prove @xmath864 .
suppose , by way of contradiction , that @xmath923 , so in particular @xmath1078 and @xmath1126 . since @xmath1123 $ ] is an even permutation we have
@xmath1129 . if @xmath1130 then @xmath859 , so by lemma [ s5.61bis ] the group @xmath849 is transitive .
using mochizuki s lemma [ s4.37a ] and the braid move @xmath705 we can transform @xmath1085 into @xmath1131 with @xmath1132 .
this contradicts the minimality of @xmath1031 . thus if @xmath923 then @xmath1133 . by lemma [ s5.61bis ]
the group @xmath855 has two domains of transitivity @xmath928 and @xmath1134 is minimal .
if @xmath1135 are the factorizations into independent cycles , then as we saw in the proof of lemma [ s6.76 ] the minimality of @xmath1031 implies that each @xmath1087 or @xmath1090 is a transposition of the type @xmath1136 where @xmath1137 .
it is clear that @xmath1138 .
let @xmath1139 where @xmath1140 means trivial extension from @xmath754 to @xmath6 .
we have @xmath1141 and the following cases may occur : a ) @xmath1142 ; b ) @xmath1143 ; c ) @xmath1144 . in case ( a ) by minimality @xmath1145 .
so @xmath1146 where @xmath1147 and @xmath1148 since @xmath1123\neq 1 $ ] .
then @xmath1123=(a_2a_1c)$ ] .
this contradicts @xmath1123=t_{1}\cdots t_{n}\in s(\sigma_1)\times s(\sigma_2)$ ] , so case ( a ) is impossible . in case
( b ) we have @xmath1149 by minimality and @xmath1150 . if @xmath1151 then @xmath1152 . so @xmath912 is non - minimal and furthermore there exists an element @xmath1153 such that @xmath1084 .
if @xmath1154 or @xmath1155 with @xmath1156 then @xmath1157 .
this contradicts lemma [ s6.76 ] .
it remains to consider the case @xmath814 .
let @xmath1158 .
suppose that @xmath1085 is a transposition .
we may assume that @xmath1159 .
we exclude case by case the possibilities for @xmath1037 .
the cases @xmath1160 , or @xmath1161 , or @xmath1162 are impossible since @xmath1123\neq 1 $ ] .
the cases @xmath1163 or @xmath1164 are impossible either since @xmath1123=s_1s_2 $ ] is not a cycle of order 3 .
it remains to exclude the possibility @xmath1165 .
indeed here @xmath1166 = ( a_1c_1)(a_1c_2)(a_2c_1)(a_1c_1)(a_1c_2)(a_2c_1)= ( a_1c_1)(a_2c_2),\ ] ] while @xmath1123 $ ] must equal @xmath1167 .
suppose that @xmath1155 .
we may assume that @xmath1168 . then @xmath1123 $ ]
might be @xmath1169 only if @xmath1163 or @xmath1164 .
the calculation above shows that in these cases @xmath1123\neq ( a_1a_2)(c_1c_2)$ ] so case ( b ) is impossible . in case ( c )
we have @xmath1170 by minimality and @xmath1117 . here
if @xmath1172 we obtain a contradiction with lemma [ s6.76 ] by the same argument we used in case ( b ) .
the cases @xmath1173 and @xmath1154 or @xmath1155 are excluded in the same way .
it remains to consider the case @xmath1174 where @xmath1175 .
let @xmath532 be the subsequence of @xmath1176 which moves elements of @xmath1177 .
replacing @xmath755 by a braid - equivalent sequence we may assume that @xmath1178 where @xmath1179 .
by main lemma [ s3.32 ] the hurwitz system @xmath1180 is braid - equivalent to @xmath1181 . if @xmath1182 then the group @xmath1183 has a domain of transitivity @xmath1184 and @xmath1185 .
this contradicts the maximality of @xmath856 .
hence @xmath1186 .
this implies that the group @xmath1024 is imprimitive which is excluded by hypothesis .
we thus excluded all possible cases with @xmath923 .
therefore @xmath1095=1 $ ] .
suppose that @xmath1151 .
then from lemma [ s6.79 ] and proposition [ s6.72 ] it follows that @xmath1026 and @xmath1029 $ ] is braid - equivalent to eq . which proves the theorem when @xmath1151 .
so , the only case that remains to be considered is @xmath1187=1 $ ] .
here we have @xmath1188 with @xmath1189 .
let @xmath1190 .
suppose , by way of contradiction , that @xmath1078 .
unless @xmath1191 where @xmath1192 we may apply the argument of lemma [ s6.79 ] and obtain a contradiction with the minimality of @xmath1031 .
we obtain a hurwitz system of the type @xmath1193 where @xmath1037 commutes with @xmath1194 .
the centralizer of @xmath1191 is @xmath1195 .
the braid move @xmath1196 transforms @xmath1037 into @xmath1197 ( cf .
thus the minimality of @xmath1031 excludes the cases @xmath1198 or @xmath1199 .
the hypothesis of primitivity of @xmath1200 excludes the cases @xmath1201 or @xmath1202 .
it thus remains to consider the cases @xmath1203 or @xmath1204 .
these are equivalent up to reordering , so it suffices to consider @xmath1203 .
we perform the following sequence of braid moves starting from @xmath1205 .
applying @xmath1053 we obtain @xmath1206 . replacing @xmath1207 by @xmath1208 and then applying @xmath705 ( cf .
theorem [ s2.27 ] ( f ) ) we obtain @xmath1209 and @xmath1210 .
so we obtain @xmath1211 . applying @xmath705 again we transform @xmath1212 into @xmath1213 .
this contradicts the minimality of @xmath1031 of the initial hurwitz system @xmath1205 .
all possible cases with @xmath1123=1,\ ; \lambda\neq 1 $ ] were excluded , so @xmath1026 .
we conclude that @xmath1214 .
the minimality of @xmath1215 implies that no independent cycle of the factorization of @xmath1037 may contain both @xmath885 and @xmath901 .
since @xmath1216 is transitive we conclude that the factorization is either @xmath1217 with @xmath1218 or @xmath1037 is a cycle of order 3 containing the two elements of @xmath1042 .
the former case is impossible since @xmath1216 is a primitive group .
we conclude that the equivalence class @xmath1219 $ ] equals @xmath1220 $ ] .
the theorem is proved .
[ s6.87 ] let @xmath3 be a smooth , projective curve of genus @xmath4 .
let @xmath796 .
if @xmath2 is prime then @xmath797 .
if @xmath2 is not prime let @xmath798 be its maximal divisor @xmath799 .
suppose that @xmath1221 or if @xmath1222 suppose that @xmath1223 .
then the connected components of @xmath11 which correspond to simple coverings with monodromy groups @xmath10 are in one - to - one correspondence with the equivalence classes of unramified coverings @xmath25 $ ] of degrees @xmath801 where @xmath802 .
s. m. natanzon , _ topology of two - dimensional coverings , and meromorphic functions on real and complex algebraic curves .
( in russian ) , trudy sem
( 1988 ) , no . 23 , 79103 .
( english translation : selecta math .
* 12 * ( 1993 ) , no . 3 , 251291 . )
vetro f. , _ irreducibility of hurwitz spaces of coverings of elliptic curve of prime degree with one point of total ramification _ , matematiche ( catania ) ( to appear ) , preprint n. 192 , may 2003 , dipartimento di matematica , universit di palermo . | graber , harris and starr proved , when @xmath0 , the irreducibility of the hurwitz space @xmath1 which parametrizes degree @xmath2 coverings of a smooth , projective curve @xmath3 of genus @xmath4 , simply branched in @xmath5 points , with full monodromy group @xmath6 .
we sharpen this result and prove that @xmath1 is irreducible if @xmath7 and in the case of elliptic @xmath3 if @xmath8 .
we extend the result to coverings simply branched in all but one point of the discriminant . fixing the ramification multiplicities over the special point
we prove that the corresponding hurwitz space is irreducible if the number of simply branched points is @xmath9 .
we study also simply branched coverings with monodromy group @xmath10 and when @xmath5 is large enough determine the corresponding connected components of @xmath11 .
our results are based on explicit calculation of the braid moves associated with the standard generators of the @xmath5-strand braid group of @xmath3 . |
during the last decade , it was shown that quantum mechanical systems have the potential for processing information more efficiently than classical systems @xcite .
however , it remains difficult to realize this potential because quantum systems are extremely sensitive to perturbations .
these perturbations come from external degrees of freedom or from the finite precision with which the systems can be realized and controlled by external fields .
this loss of quantum information to the environment is called decoherence @xcite .
different results show that a state is more sensitive to decoherence as the number of qubits increases @xcite .
this is also manifested by the impossibility to time reverse a quantum evolution when an initially localized excitation spreads over a system @xcite .
as the information is distributed over a increasing number of qubits , the evolution becomes more sensitive to the perturbation @xcite . in a similar vein
, this sensitivity with the systems size limits the distance over which one can transfer information or analogously limits the number of qubits that one can control reliably @xcite .
this is manifested as a localization effect for the quantum information @xcite . in order to overcome these limitations for allowing quantum information processing with large number of qubits , methods for reducing
the decoherence effects have to be developed .
one can tackle the problem by correcting the errors generated by the perturbations , but this is only possible if the perturbation is small enough to keep the quantumness of the system @xcite .
therefore , one needs first to reduce the perturbation effects .
the pioneering strategies for reducing decoherence were introduced in the nuclear magnetic resonance ( nmr ) community , in particular by erwin hahn who showed that inverting a spin-1/2 system ( a qubit ) corresponds to an effective change of the sign of the perturbation hamiltonian and therefore generates a time reversal of the corresponding evolution @xcite .
this leads to the formation of an echo that later was formalized as a loschmidt echo @xcite .
these manipulations were extended to the so - called decoupling methods @xcite , which disconnect effectively the environment . in the context of this review , we describe the environment as a spin - bath , without loss of generality . considering spins 1/2 as qubits , two different types of decoupling methods can be distinguished . in the first case ,
the qubit system is well distinguished from the environment .
its energy level splitting differs significantly from that of the bath . as a result
, the coupling between them is much smaller than the difference of their energy level splittings .
the interaction can then always be approximated by an ising - type ( @xmath0 ) interaction , which causes dephasing of the system qubit but no qubit flips .
the decoupling methods required for these cases are called heteronuclear decoupling within the nmr community and they can involve manipulation only on the spin - system @xcite or only at the environment @xcite .
in the second case , the system and the environment have similar energy level splittings .
this is the case of a homonuclear system where the general form of the coupling must be retained and it can induce flips of the system qubit as well as dephasing . in this case , decoupling will generally affect the complete system plus `` bath '' @xcite .
in this review we focus on decoupling the system from the environment by applying control pulses only to the system . during the last years
this technique has gathered a lot of interest because it requires relatively modest resources : it requires no overhead of information encoding , measurements or feedback .
the method is known as dynamical decoupling ( dd ) . since its initial introduction @xcite ,
a lot of effort has been invested to find sequences with improved error suppression @xcite .
the optimal design of dd sequences depends of the different sources of errors that have to be eliminated .
the first to be considered is the nature of the se interaction , i.e. , a pure dephasing when only the @xmath1-spin operator @xmath2 is present , a pure spin - flip interaction ( @xmath3 and/or @xmath4 are present ) or a general interaction with @xmath3 , @xmath4 and @xmath2 .
sequences like carr - purcell ( cp ) and carr - purcell - meiboom - gill ( cpmg ) , which use rotations around a single axis , are useful when only two operators of @xmath3 , @xmath4 and @xmath2 are present , but in order to fight against a general interaction , pulses along different spatial direction have to be applied . the shortest sequence that fulfills this condition is the xy-4 sequence @xcite .
an actual implementation must take into account , in addition to the above issues , the effect of pulse imperfections @xcite .
fighting the effect of pulse errors was in fact the original motivation for the development of the xy-4 sequence @xcite .
another experimental consideration is the amount of power deposited in the system , which often must be kept small to avoid heating effect or damage to the sample .
dd technique is becoming an important tool for quantum information processing @xcite as well as in spectroscopy @xcite and imaging @xcite .
in most cases , the goal is to preserve a given input state , but it may also be combined with gate operations @xcite . in many cases ,
experimental results show that one of the main limitations for improving the decoupling efficiency are the non - ideal properties of the decoupling pulses @xcite .
decoherence effects due to the environment can in principle be reduced by shortening the cycle time @xmath5 . however , in reality the pulse lengths are finite and the dd cycle time is limited by the available hardware . it is therefore important to find the best finite time sequences for decoupling @xcite .
furthermore , the pulses do not only have finite lengths , they also do not implement perfect rotations .
thus , increasing the number of pulses can result in a large overall error that destroys the qubit coherence instead of preserving it . as a result
, the performance of the decoupling sequence may have an optimum at a finite cycle time @xcite . in this review ,
we summarize the different dd strategies for fighting the effect of pulse imperfections and we show that they must be considered in the designing of useful dd sequences . the paper is structured as follows : in section ii we give some basics of dynamical decoupling , in section iii we introduce the effects of pulse imperfections and in section iv we describe the strategies to fight against the imperfections effects . in the last section
we draw some conclusions and perspectives .
we consider a single qubit @xmath6 as the system that is coupled to the bath . in a resonantly rotating frame of reference @xcite ,
the free evolution hamiltonian is @xmath7 where @xmath8 is the environment hamiltonian and @xmath9 is a general interaction between the system and the environment .
@xmath10 are operators of the environment and @xmath11 the se coupling strength .
the index @xmath12 runs over all modes of the environment .
dephasing is due to an interaction that affects the @xmath13 component of the spin - system operator , and spin - flips are done trough the @xmath14 and/or @xmath15 operators .
a heteronuclear spin - spin interaction involves a pure dephasing interaction .
this type of interaction that is naturally encountered when the system can be distinguished from the environment can be found in a wide range of solid - state spin systems , as for example nuclear spin systems in nmr @xcite , electron spins in diamonds @xcite , electron spins in quantum dots @xcite , donors in silicon @xcite , etc . in other cases when the system and environment have similar energies ,
the se interaction can include terms along the @xmath14 , @xmath15 and @xmath13 axis .
dd is achieved by iteratively applying to the system a series of stroboscopic control pulses in cycles of period @xmath5 @xcite . over that period
, the time - averaged se interaction can be described by an averaged or effective hamiltonian @xcite .
the goal of dd is the elimination of the effective se interaction .
this can be seen by looking at hahn s pioneering spin - echo experiment @xcite ( see fig .
[ ddseqs]b ) .
it is based on the application of a @xmath16-pulse to the spin system at a time @xmath17 after the spins were left to evolve in the magnetic field .
this pulse effectively changes the sign of the system - environment ( se ) interaction , in this case the zeeman interaction with the magnetic field . letting the system to evolve for a refocusing period or time reversed evolution during the same duration
@xmath17 generates the echo . if the magnetic field is static , the dynamics is completely reversed and the initial state of the spin recovered
this is observed as the echo .
however if the magnetic field fluctuates , its effect can not be reversed completely .
thus , the echo amplitude decays as a function of the refocusing time @xcite .
this decay contains information about the time - dependence of the environment . dynamical decoupling pulse sequences .
the empty and solid rectangles represent @xmath18 and @xmath19 pulses , respectively and @xmath20 represents the number of iterations of the cycle .
( a ) initial state preparation .
( b ) hahn spin - echo sequence .
( c ) cpmg sequence .
( d ) cdd sequence of order @xmath21 , cdd@xmath22 = c@xmath22 and c@xmath23.,title="fig:",width=321 ] to reduce the decay rate of the echo due to a time - dependent environment , carr and purcell introduced a variant of the hahn - echo sequence , where the single @xmath16-pulse is replaced by a series of pulses separated by intervals of duration @xmath17 @xcite .
this cp sequence reduces the changes induced by the environment if the pulse intervals are shorter than the correlation time of the environment .
however , as the number of pulses increases , pulse errors tend to accumulate .
their combined effect can destroy the state of the system , rather than preserving it against the effect of the environment .
this was noticed by meiboom and gill who proposed a modification of the cp sequence for compensating pulse errors , the cpmg sequence @xcite .
cp and cpmg sequences are useful only when two of the spin operators @xmath3 , @xmath4 and @xmath2 are affected by the environment .
they can be written as @xmath24 where @xmath25 is a free evolution operator and @xmath26 is a @xmath16 pulse around the @xmath27 axis ( and the analogous for @xmath28 ) .
the difference between the cp and cpmg sequence is the orientation of the rotation axis with respect to the initial condition . for applications in quantum information processing ,
this distinction is not relevant , since all operations have to be independent of the initial condition .
cp or cpmg is the shortest sequence of pulses for decoupling a se interaction that affects only two of the spin components @xmath2 , @xmath4 and @xmath2 @xcite . usually the average hamiltonian generated by dd sequences can be described by a series expansion , such as the magnus expansion@xcite .
all the higher - order terms in this expansion describe imperfections , which reduce the fidelity of the sequence and should be eliminated .
improving the dd performance is therefore related to reducing the contribution of higher order terms .
this is closely related to efforts for developing better decoupling sequences for nmr @xcite . for quantum information processing ( qip ) , this lead to the design of sequences that make dd more effective , such as concatenated dynamical decoupling @xcite .
an important innovation was due to g. uhrig @xcite , who proposed a sequence with non - equidistant pulse spacings , while all the standard sequences like cpmg are based on equidistant pulses .
the udd sequence is defined by @xmath29 where the delays @xmath30 are determined by the positions @xmath31\ ] ] of the pulses with @xmath32 and @xmath33 .
the lowest nontrivial order is equal to the cpmg sequence , udd@xmath34 cpmg .
the effect of the non equidistant pulses can be discussed in the context of filter theory : dd can be considered as an environmental noise filter , where the distribution of pulses generates different filter shapes as a function of the frequencies .
the overlap of this filter function with the spectral distribution of the environmental noise determines the decoherence rate @xcite .
analogously , the filter shapes can be connected to diffraction patterns induced by interferences in the time domain @xcite .
the uhrig dd sequence was shown theoretically to be the optimal sequence for reducing low frequency noise @xcite .
this prediction was confirmed experimentally @xcite .
however , it appears that non - equidistant sequences perform better only for particular noise spectral densities that increase for higher frequencies and have a strong cut off . in usual case , where the spectral density decrease smoothly with the frequency , as usually happens with nmr spin baths , equidistant sequences were predicted @xcite and demonstrated @xciteto be the best option @xcite . this filter function description can be traced back to previous nmr approaches @xcite and to work on universal dynamical control @xcite .
choosing the times for the pulses leads to a variety of sequences that can be optimized according to the spectral density of the bath @xcite . if the se interaction is a pure dephasing one , it is sufficient to apply pulses in one direction .
however because every experimental setup has finite precision , pulse errors create an effective general se interaction @xcite . in this case
, it was shown that sequences that are designed for general se interaction perform better than 1d sequences @xcite .
if the system - environment interaction includes all three components of the system spin operator , decoupling can only be achieved if the sequence includes rotations around at least two different axes .
the first decoupling sequence of this type is the xy-4 sequence , which alternates rotations around the x- and y - axes ( see fig .
[ ddseqs ] d , @xmath35 ) .
this sequence was initially used to eliminate the effect of pulse errors in the cp and cpmg sequences @xcite .
it is also the shortest sequence for dd for general se interactions @xcite . in quantum information processing , where we consider the initial state to be unknown ,
the cp and cpmg sequences , which correspond to a train of identical @xmath16-pulses , are identical @xcite .
however , their effect on the quantum state depends on the ( generally unknown ) initial condition : if the initial condition is oriented along the rotation axis of the pulses , flip angle errors of the first pulse are refocused by the second pulse .
however , for components perpendicular to the rotation axis , the pulse errors of all pulses add and cause rapid decay of the coherence , even in the absence of system - environment interactions @xcite .
this motivated the development of the xy-4 sequence .
in addition , pulse imperfections convert an ising - type system environment interaction into an effective general se interaction @xcite , which is not eliminated by the cp / cpmg sequence , but is partially eliminated by the xy-4 sequence . in the qip community , the xy-4 sequence is usually referred to as periodic dynamical decoupling ( pdd ) .
the xy-4 sequence is also the building block for concatenated dd ( cdd ) sequences that improve the decoupling efficiency @xcite .
the cdd scheme recursively concatenates lower order sequences to increase the decoupling power .
the cdd evolution operator for its original version for a recursion order of @xmath21 is given by @xmath36 where @xmath37 and @xmath38=xy-4 .
figure [ ddseqs ] shows a general scheme for this process .
each level of concatenation reduces the norm of the first non - vanishing order term of the magnus expansion of the previous level , provided that the norm was small enough to begin with @xcite .
this reduction comes at the expense of an extension of the cycle time by a factor of four .
if the delays between the pulses are allowed to be non - equidistant like in udd , it becomes possible to create hybrid sequences , such as cudd @xcite and qdd @xcite .
since the precision of any real operation is finite , the control fields used for decoupling introduce errors . depending on the sequence , these errors can accumulate .
if the number of pulses is large and the sequence is not properly designed , the accumulated pulse errors can lead to severe loss of coherence than the effect of the environment . designing effective decoupling sequences that suppress environmental effects without degrading the system , even if the control fields have errors ,
requires a careful analysis of the relevant errors and appropriate strategies for combining rotations in such a way that the errors cancel rather than accumulate .
one example of a non - ideal control pulse is its finite duration , which implies a minimum achievable cycle time .
the effects introduced by finite pulse lengths have been considered in different theoretical works @xcite .
these works predict that high order cdd or udd sequences in general lose their advantages when the delays between pulses or pulse length are strongly constrained .
while the limitation on the cycle reduces the maximal achievable dd performance , pulse errors can be even more destructive . in most cases ,
the dominant cause of errors is a deviation between the ideal and the actual amplitude the of control fields .
the result of this amplitude error is that the rotation angle experimentally implemented deviates from @xmath16 , typically by a few percent .
another important error occurs when the control field is not applied in resonance with the transition of the qubit .
this off - resonant effect produces a rotation in which the flip angle and the rotation axis deviate from the ideal ones .
[ cols="^,^,^ " , ] an example of the destructive effects of pulse imperfection is illustrated in the left panel of figure [ chim ] .
here , we measured the magnetization decay of the @xmath39c nuclear spins during two different dd sequences .
the sample used for this experiment was polycrystalline adamantane @xcite .
the dephasing of the nuclear spins originates from the interaction with an environment consisting of @xmath40h nuclear spins and can be considered as a pure dephasing process .
the first sequence considered in the figure is cpmg . in this case
we can observe that the decay of the magnetization is imperceptibly slow when the system is initially oriented parallel to the rotation axis of the pulse ( longitudinal state ) . as we discuss below , this is an indication that the pulse errors have no effect on this initial state .
in contrast , for a transverse initial state , the errors of the individual pulses accumulate and lead to a rapid decay , as shown in fig .
a similar behavior is found for the udd sequence @xcite .
the second dd sequence considered in fig . [ chim ] is the xy-4 sequence , which consists of pulses applied along the x and y axes .
the alternating phases of the pulses results in a partial cancellation of pulse errors , independent of the initial condition @xcite . as a result ,
the performance of this sequence is much more symmetric with respect to the initial state in the xy - plane and the average decay times are significantly longer @xcite . in the context of quantum information processing ,
it is important that the performance of gate operations be independent of the initial conditions ( which typically are unknown ) .
a common choice for quantifying the performance of a general quantum operation is then the fidelity @xmath41 @xcite : @xmath42 here , @xmath43 is the target propagator for the process and @xmath44 the actual propagator . for the present situation ,
where the goal is a quantum memory , the target propagator is the identity operation @xmath45 .
we can not assume that the actual propagators are unitary .
we therefore write the process as @xmath46 where @xmath47 and @xmath48 are the density matrices at the beginning and end of the process .
the operators @xmath49 must form a basis .
for the present case , we choose them as @xmath50 .
the ideal and actual processes can therefore be quantified by the matrix elements @xmath51 . for the target evolution ,
the @xmath52-matrix is @xmath53 the matrix elements for the actual process are determined experimentally by quantum process tomography @xcite .
we use them to calculate the process fidelity from eq .
( [ e : fid ] ) . in figure [ chim ] , we compare the real part of the experimental @xmath52 matrices for the two sequences .
after one cycle , the process matrices for both sequences represent a quantum operation that is close to the identity operation .
the fidelity between the experimental matrices and the ideal matrix ( [ chiid ] ) is @xmath54 and @xmath55 for the cpmg and xy-4 cycles , respectively .
measured over 40 cycles , the process matrices of the two sequences differ significantly from the identity operation but also from each order . for the xy-4 sequence ,
the non - vanishing elements are @xmath56 and @xmath57 , while for cpmg the non - vanishing elements are @xmath56 and @xmath58 .
the two matrices represent therefore qualitatively different processes .
the xy-4 sequence destroys all transversal magnetization , in this case the resulting density matrix is @xmath59 this corresponds to a projection of the density operator onto the @xmath13-axis , i.e. to a complete dephasing of the transverse components in the @xmath60-plane .
the cpmg sequence , conversely , projects the density operator onto the @xmath14-axis : @xmath61 the cpmg is a spin lock @xcite sequence that allows retains magnetization in the @xmath14 direction but destroys all components perpendicular to it . since a real experimental implementation always generates a distribution of control field amplitudes , spins at different positions precess with different rates around the direction of the rf field . as a result , the perpendicular components become completely randomized after a sufficiently large overall flip angle as shown in the left panel of fig .
[ chim ] .
the performance of experimentally accessible dd sequences is limited by the pulse errors @xcite . in many situations ,
the dominating error contributions are flip angle and offset errors . in the next section
we show different strategies to make decoupling sequences robust against such errors .
we have to make decoupling insensitive to pulse imperfections .
different possibilities for generating high - fidelity sequences have been proposed in the context of quantum information processing @xcite . here ,
we discuss two possible approaches : first we show that it is possible to replace individual refocusing pulses by compensated pulses that implement very precise inversions , and then we discuss sequences that are inherently robust , i.e. insensitive to the imperfections of the individual pulses . the simplest approach to make a sequence robust is by replacing every standard pulse by a robust composite pulse @xcite .
the composite pulses are sequences of consecutive pulses designed to be robust against various classes of imperfections generating therefore rotations that are close to the ideal rotation even in the presence errors . particularly useful for quantum information applications are those composite pulses that produce compensated rotations for any initial condition , denominated in the nmr literature as class - a pulses @xcite .
recent experiments have successfully used composite pulses to demonstrate the resulting increase of the performance of different dd sequences @xcite .
these works have implemented the composite pulse defined as : @xmath62 which is equivalent to a robust @xmath16 rotation around the axis defined by @xmath63 followed by a @xmath64 rotation around the @xmath13 axis @xcite . for cyclic sequences , which always consist of even numbers of @xmath16 rotations , the effect of the additional @xmath13 rotation vanishes if the flip angle errors are sufficiently small .
a comparison between standard sequences ( not using robust pulses ) against sequences with robust pulses has been reported in @xcite .
it was observed that robust pulses improve the performance at high duty cycles .
however , for low duty cycles , standard sequences perform better .
this is due to the shorter cycle time of the standard sequence if constant duty cycles are compared .
thus , if the objective is only to preserve a quantum state , the best performance is obtained at high duty cycles , using robust pulses .
the best sequences that are suitable for parallel application of quantum gate operations are the self - correcting sequences discussed below .
the composite pulses are usually designed to correct flip angle errors and offset errors . for compensating the effects introduced by the finite length of pulses ,
some theoretical works have proposed that a finite pulse could be approximated as an instantaneous one by using an appropriate shaped pulse @xcite .
these works has provided analytical and numerical evidence that a careful shape design can strongly affects the performance of decoupling sequences .
an alternative to the use of composite pulses consists in making the decoupling sequences fault - tolerant without compensating the error of each pulse , but by designing them in such a way that the error introduced by one pulse is compensated by the other pulses of the cycle @xcite . a straightforward strategy for designing improved sequences
consists in concatenating one basic building block cycle into a longer and robust cycle . the xy-4 cycle is often chosen as the building block for constructing self - correcting sequences .
this cycle is the shortest dd sequence that cancels the zero - order average hamiltonian for a general se interaction and also has the advantage of being partially robust to pulse imperfections . in the spectroscopy and quantum computing communities , two versions of the xy-4 sequence
are used @xcite .
the basic cycle originally introduced in nmr shows reflection symmetry with respect to the center of the cycle .
in contrast to that , the sequence used in the quantum information community is time asymmetric .
one consequence of this small difference is that in the symmetric version , the echoes are formed in the center of the windows between any two pulses , while in the case of asymmetric cycles , the echoes coincide with every second pulse , as shown in figure [ ecos ] .
the separation in time between the echoes is therefore twice as long in this case .
if the environment is not static , the larger separation of the echoes leads to a faster decay of the echo amplitude @xcite .
if a robust pulse only contains @xmath16 rotations , as in the case of ( [ kpul ] ) , it is also possible to convert such a composite pulse into a decoupling cycle by inserting delays between the individual @xmath16 rotations .
this approach has been used in @xcite to build an self - correcting sequence .
the basic cycle is then @xmath65 the self - correcting sequence is created by combining 5-pulse blocks shifted in phase by @xmath66 , such as [ kdd@xmath67 - kdd@xmath68@xmath69 , where the lower index gives the overall phase of the block . the cyclic repetition of these 20 pulses is referred to as the kdd sequence @xcite . in fig .
[ fidelity ] , we show how strongly errors in the flip angles of individual pulses affect the fidelity of the pulse sequence . neglecting the effect of the environment , we calculate the fidelity @xmath41 after the application of 20 pulses as a function of the flip angle error .
the figure compares the fidelities for the cpmg , xy-4 and kdd cycles . for the cpmg sequence
, the fidelity drops to < 95% if the flip angle error exceeds @xmath70 2 % . for the xy-4 sequence ,
this bandwidth increases to @xmath70 10 % and for kdd to @xmath70 30 % .
kdd and xy-4 are obviously much less susceptible to pulse imperfections than cpmg .
the low fidelities observed for cpmg is experimentally manifested by the fast decay of the transverse components , such as @xmath71 in fig .
[ chim ] . schematic representation of time symmetric xy-4(s ) and asymmetric xy-4(a ) . ,
width=396 ] simulation of fidelity as a function of the flip angle error for kdd , xy-4 and cpmg cycles . ,
title="fig:",width=472 ] every decoupling sequence contains unwanted terms in the average hamiltonian .
they can be reduced by combining different versions of the basic cycles in such a way that some of the error terms cancel .
two different versions of this procedure have been used : the basic cycles can be applied subsequently @xcite or one cycle can be inserted into the delays of another cycle @xcite .
the first approach was introduced in nmr , e.g. for designing high - performance homonuclear decoupling sequences @xcite or in high - resolution heteronuclear decoupling @xcite .
examples of dd sequences that can be constructed using this approach are xy-8 and xy-16 sequences @xcite . here the xy-8 is created combining a xy-4 cycle with its time - reversed image while xy-16 is created combining the xy-8 with its phase - shifted copy .
the second approach is the concatenation scheme proposed by khodjasteh and lidar @xcite .
it generates the cdd sequence of order @xmath72 by inserting cdd@xmath22 cycles into the delays of the xy-4 sequence ( see figure [ ddseqs ] ) .
ideally , each level of concatenation improves the decoupling performance and the tolerance to pulse imperfections ; in practice , higher order sequences do not always perform better . it has been theoretically predicted @xcite and later observed experimentally that the finite duration of the pulses and constrained delays between pulses result in the existence of optimal levels of concatenation @xcite . in figure
[ sfseqs ] , we demonstrate how the well - designed combination of basic cycles can lead to extended cycles with better error compensation . here , we consider as the leading experimental imperfections deviations of the amplitude and frequency of the pulse .
neglecting the effect of the environment , we calculate the fidelity @xmath41 after applying 1680 pulses to the system as a function of the two error parameters .
each panel contains the color - coded fidelity for one of six different decoupling sequences . in the top panels we can clearly see the improvement in the error tolerance due to the cdd scheme of concatenation .
panels 1 , 4 and 5 show the same result for the sequential concatenation scheme , where only two cycles are combined at each step : concatenation of the xy-4 cycle with its time - inverted and phase - shifted copies forms the xy-8 and xy-16 sequences .
the 16-pulse xy-16 cycle is significantly more robust than the 84-pulse cdd@xmath73 cycle .
the best performance is achieved by the kdd sequence , whose cycle consists of 20 pulses .
error tolerance of different self - correcting sequences .
the upper row shows the calculated fidelity f for cdd sequences , while the lower row shows the results for xy-8 , xy-16 and kdd sequences .
each panel shows the fidelity after 1680 pulses as a function of flip - angle error and offset errors .
the regions where the fidelity is lower than 0.95 are shown in white.,title="fig:",width=548 ] in fig .
[ kddfig ] , we compare the experimental performance of different self - correcting sequences .
the performance of the cdd sequences always saturates or decreases with increasing duty cycle under the current experimental conditions @xcite
. however , instead of saturating , the relaxation time for the kdd sequence continues to increase , as in the case of sequences with robust pulses @xcite .
the kdd sequence combines the useful properties of robust sequences with those of sequences of robust pulses and can thus be used for both quantum computing and state preservation .
experimental decoherence times for different compensated dd sequences as a function of the duty cycle.,title="fig:",width=396 ] the symmetry of the basic building blocks has a key role in determining the performance of the concatenated higher order sequences .
two sequences constructed according to the same rules from a basic block have different propagators if the basic blocks are symmetric or not @xcite .
if we concatenate four xy-4 cycles to the xy-16 sequence , e.g. , we obtain new cycles , which are time - symmetric , independent of which version of the xy-4 sequence was used for the building blocks .
although all the odd order terms vanish in the average hamiltonians of both versions , the even order terms of the sequences that are built from asymmetric blocks contain additional unwanted terms @xcite .
the different behavior of sequences consisting of symmetric vs. asymmetric blocks is illustrated in figure [ fidelxy ] .
if we start from the symmetric form of xy-4 , the resulting xy-16 sequence shows much better performance than the sequence using the asymmetric xy-4 as the building block .
analogous results were obtained for the two versions of the xy-8 sequence @xcite .
earlier experiments showed two different contributions to the overall fidelity loss @xcite : a precession around the @xmath13-axis , which can be attributed to the combined effect of flip - angle errors and an overall reduction of the amplitude , which results from the system - environment interaction .
the combination of precession and reduction of amplitude is illustrated in right panel of fig .
[ fidelxy ] . in this figure
we show the xy - components of the magnetization at different times during the xy-16 sequence . if the xy-16 sequence is built by the asymmetric form of xy-4 a distinct precession around the @xmath13-axis
is observed .
this causes a deviation from the desired evolution and reduces therefore the fidelity of the process .
however , for the sequence consisting of symmetric blocks , the precession is negligible .
these results suggest that pulse errors are better compensated by concatenating symmetric building blocks . experimental fidelity decay for cdd@xmath74 built from symmetric and asymmetric blocks .
the delay @xmath75 between the pulses is constant and identical for both sequences .
, width=472 ] the same concept can also be applied to cdd sequences .
the conventional concatenation scheme of eq .
( [ cdd ] ) uses asymmetric building blocks and is not compatible with the symmetric version of xy-4 .
a new concatenation scheme was therefore proposed in @xcite . in this scheme , the symmetrized version of cdd
is constructed as @xmath76 ^ 2 .\end{aligned}\ ] ] in fig .
[ fcdd ] we compare the process fidelities for the two versions of the cdd@xmath74 sequence . as in the case of xy sequences , clearly , the symmetrized version , cdd@xmath74(s ) , shows a significantly improved performance , compared to the standard cdd@xmath74(a ) version . in ref .
@xcite , it was experimentally observed that the performance of all dd sequences based on symmetric building blocks is better or equal to that of sequences using non - symmetric building blocks .
this behavior is consistent with general arguments based on average hamiltonian theory @xcite .
dynamical decoupling is becoming a standard technique for preserving the coherence of quantum mechanical systems , which does not need control over the environmental degrees of freedom .
the technique aims to reduce decoherence rates by attenuating the system - environment interaction with a periodic sequence of @xmath16 pulses applied to the qubits .
the pioneering strategies for decoupling were introduced in the context of nmr spectroscopy @xcite . since then
, many different decoupling sequences have been developed in the context of nmr @xcite or quantum information processing @xcite .
generally , we can divide the dd sequences in two groups : ( i ) sequences that involve pulses in a single spatial direction and ( ii ) sequences that contain pulses in different spatial directions . the type ( i )
sequences are strongly sensitive to pulse errors and are only capable of suppressing the effects of a purely dephasing environment or pure spin - flip interaction . examples of such dd sequences are cpmg and udd .
the second group can suppress a general se interaction and usually exhibits better tolerance to experimental imperfections .
example of such sequences are the xy - family ( xy-4 , xy-8 and xy-16 ) , the cdd sequences and the kdd sequence .
recent experiments have successfully implemented dd methods and demonstrated the resulting increase of the coherence times by several orders of magnitude @xcite .
these works also showed that the main limitation to the reduction of the decay rates are the imperfections of the pulse .
two approaches have been used to correct this .
the first approach replaces the inversion pulses by robust composite pulses @xcite , which generate rotations that are close to the target value even in the presence of pulse errors . in this case
, the pulses are corrected individually .
the second approach consists in designing the decoupling sequences in such a way that the error introduced by one pulse is compensated by the other pulses , without compensating the error of each pulse individually .
the properties of basic decoupling cycles can be further improved by concatenating basic cycles into longer and more robust cycles .
the concatenation can be made either by combining symmetry - related copies of a basic cycle subsequently @xcite ( resulting in the xy-8 and xy-16 sequences ) or by inserting the basic cycle into the delays of another cycle @xcite ( cdd sequences ) .
the time reversal symmetry of the basic building blocks is a useful criterion for minimizing error contributions .
it has been demonstrated that the sequences built from symmetric building blocks often perform better and never worse than sequences built from non - symmetric blocks @xcite .
this is a significant advantage , considering that the complexity of the sequences based on symmetric or asymmetric blocks are identical .
earlier experiments @xcite showed that the best sequences that are suitable for parallel application of quantum gate operations are the symmetric self - correcting sequences .
however , as the delay between pulses decreases , sequences with robust pulses perform better .
thus , if the objective is only to preserve a quantum state , the best performance is achieved by using robust pulses to correct pulse errors . on the other hand ,
the kdd sequence introduced in @xcite combines the useful properties of self - correcting sequence with those of robust pulses and can thus be used for both quantum memory and quantum computing . during the last years
, many advances have been achieved . however , for the application of the technique in real quantum devices , further studies will certainly be required .
so far , most work has focused on single qubit systems . in the future , more experimental tests will be needed with multi - qubit systems . in the field of quantum computation
, another important development may result from the combination of dynamical decoupling sequences with those techniques used to implement robust quantum gates @xcite .
since dd does not require auxiliary qubits or measurements , it can be used as an economical alternative to complement quantum error correction .
some theoretical works @xcite proposed methods for combining the two methods but no experimental test was carried out to date .
future research on dynamical decoupling will also focus on applications outside of quantum information processing .
recent experiments have applied dd pulse sequences , for example , to probe the noise spectrum directly @xcite and detect weak magnetic fields @xcite . | quantum computers , which process information encoded in quantum mechanical systems , hold the potential to solve some of the hardest computational problems . a substantial obstacle for the further development of quantum computers
is the fact that the life time of quantum information is usually too short to allow practical computation . a promising method to increase the life time , known as dynamical decoupling , consists of applying a periodic series of inversion pulses to the quantum bits . in the present review ,
we give an overview of this technique and compare different pulse sequences proposed earlier .
we show that pulse imperfections , which are always present in experimental implementations , limit the performance of dynamical decoupling .
the loss of coherence due to the accumulation of pulse errors can even exceed the perturbation from the environment .
this effect can be largely eliminated by a judicious design of pulses and sequences .
the corresponding sequences are largely immune to pulse imperfections and provide an increase of the coherence time of the system by several orders of magnitude . |
turbulent thermal convection plays a key role in many of the phenomena associated with geophysical and astrophysical fluid dynamics @xcite as well as providing a well - posed problem for the study of fundamental fluid dynamics @xcite . in several important examples including oceanic deep convection @xcite and convection in stars @xcite and giant planets @xcite ,
the effects of rotation are critical in determining the nature of the fluid motion .
rotation also provides an additional parameter for understanding the origin of heat transport scaling in turbulent convection , a topic of tremendous experimental activity in recent years @xcite . in comparison ,
the research efforts applied to rotating turbulent convection have been rather modest arising from the pioneering theoretical work of chandrasekhar @xcite .
experimental measurements of heat transport in rotating convection include the seminal work of rossby @xcite and later studies that also had qualitative flow visualization @xcite .
numerical simulations have also had significant impact @xcite .
here we consider both rotating and non - rotating convection and provide new insights into heat transport scaling of rotating convective turbulence .
a short report of some aspects of this work appeared previously @xcite , and further studies of velocity fields in rotating convection motivated by this work were also published @xcite .
rotating rayleigh - bnard convection can be characterized by three dimensionless parameters : the rayleigh number ra which is a measure of buoyant forcing , the taylor number ta which measures the effect of the rotational coriolis force , and the prandtl number @xmath0 which determines the dominant nonlinearity in convection .
these parameters are defined by : @xmath10 where @xmath11 is the acceleration of gravity , @xmath12 is the thermal expansion coefficient , @xmath13 is the temperature difference across the fluid layer of height @xmath14 , @xmath15 is the kinematic viscosity , @xmath16 is the thermal diffusivity , and @xmath6 is the physical angular rotation rate . the dimensionless rotation frequency @xmath17 is sometimes used in place of @xmath6 or ta
. properties of thermal turbulence can also be affected by the cell geometry characterized by the ratio of a lateral length to a vertical length . for our square geometry
, we define the cell aspect ratio as @xmath18 where @xmath19 is the lateral size of the cell .
although ra , ta , @xmath0 and @xmath20 completely define the parameter space of rotating convection , the behavior of different quantities such as heat transport is complicated when one parameter is varied while keeping the others constant .
for example , as ra is increased at fixed ta , the relative influence of buoyancy and rotation changes , making an evaluation of the influence of rotation alone difficult . to ease this problem , it is useful to define @xcite an additional parameter , the convective rossby number ro @xmath21 which is a measure of a characteristic buoyant velocity to a rotational velocity .
this definition is equivalent to those used previously @xcite and is closely related to other definitions of convective rossby number @xcite . roughly speaking
, the border between rotation - dominated and buoyancy - dominated flows should be approximated by the condition ro = 1 .
the quantity of interest here is turbulent heat transport as measured by the nusselt number nu , the total heat transported by convection normalized by the heat transported by molecular diffusion alone . to appreciate the influence of rotation on nu
, it is important to understand the dependence of nu on ra without rotation .
the investigation of non - rotating convection has been extensive over almost 40 years with early work focused on classical theories @xcite that predicted a power law relationship of the form nu = a ra@xmath22 with @xmath23 .
later measurements , particularly those in helium gas , suggested a value @xmath24 with theory and early numerical simulations providing a solid basis for such a law .
a detailed review of these results was presented by siggia @xcite .
an extension @xcite of the competing kinetic and thermal boundary layer theory @xcite that included an expanded analysis of competing boundary and bulk dissipation processes produced a phase diagram with crossover effects between different regions .
such an approach suggested a form nu = a ra@xmath25 + b ra@xmath26 with specific predictions for the coefficients derived from fitting a few data sets in each region . in this
latter regard , high precision experimental data for room temperature fluids @xcite have been extremely valuable in elucidating differences between a single power law description and one involving two power laws with fixed coefficients . the measurements presented here are over a modest range of ra @xmath27 , and thus can not distinguish between these two forms .
we compare our results with a broad range of measurements of heat transport of non - rotating convection as a benchmark for considering the effects of rotation on turbulent heat transport .
the effects of rotation on convection , especially on heat transport , might be expected to be substantial given that rotation profoundly changes the nature of boundary layer instability and modifies the length scales over which motions occur . whereas thermal plumes are formed in long sheets and are swept across the cell by mean flow
, rotation spins up these plumes into intense vortical structures .
furthermore , rotation is known to shorten the linear length scale dramatically as rotation is increased @xcite .
additional ingredients introduced by rotation are the ekman pumping / suction imposed by the differential rotation of the boundary and the interior flow and the dynamical constraints imposed by the taylor - proudman theorem for strongly rotating flows . despite these interesting factors ,
previous heat transport measurements have not been well understood for a number of reasons .
rossby , in his seminal paper on rotating convection @xcite , reported comprehensive heat transport for water and for mercury as function of ra and ta with emphasis on the regions close to onset and of moderate ra ( @xmath28 ) .
his measurements , as well as later measurements in helium @xcite , quantitatively showed that the convective onset was below the theoretical prediction of linear stability analysis @xcite .
this reduction in critical rayleigh number was attributed to a transition to azimuthally - periodic modes localized near the wall @xcite but neither of the experiments had flow visualization capabilities .
the existence of such wall states was later confirmed @xcite using shadowgraph flow visualization but rather than being stationary the sidewall states were observed to precess in the rotating frame counter to the direction of rotation .
this resolved one difficulty with the data set of rossby .
other experiments @xcite and numerical simulations @xcite of rotating convection involve an open upper fluid surface where one can visualize the development of convective structures and the interaction of vortices and characterize some of the statistics of the temperature and velocity fields .
these experiments are , however , not amenable to the measurement of accurate heat transport .
the heat transport experiments @xcite that motivated this work used a water in a cylindrical cell with top and bottom rigid boundaries , and measurements were made to higher ra ( @xmath29 ) than rossby .
the normalized heat transport nu at constant rotation rate appeared , however , to asymptote to the non - rotating result at high ra .
this combination of rather small ra and an apparently non power - law scaling was confusing .
numerical simulations @xcite showed , however , that nu scaled approximately as the 2/7 power for a fixed convective rossby number with no - slip top and bottom boundary conditions . to test this prediction and to further characterize heat transport as a function of rotation
, it was necessary to extend the heat transport measurements to higher ra than in @xcite .
single point measurements of temperature made in the same cell as the one used here will be presented elsewhere .
this paper is organized as follows .
the experimental apparatus and procedures are decribed in sec .
[ sec : setup ] .
the heat transport results of non - rotating and rotating turbulence are presented in sec .
[ sec : nr ] and sec .
[ sec : r ] , repectively .
[ sec : con ] summarizes the paper .
the experimental apparatus , shown schematically in fig . [ fig : setup ] , was an improved version of the one used previously in studies of rotating convection @xcite .
the top plate temperature of the cell was controlled by water flow which was distributed evenly by a set of 12 turrets divided into two groups pointing to 1/3 and 2/3 of the radius , respectively .
the water flow was temperature - controlled by a refrigerator - circulator and then fed into the rotating frame through a water slip connection .
the flow was further temperature regulated by a feedback temperature control unit which maintained the top - plate temperature constant with r.m.s .
fluctuations of less than 1 mk . a film heater attached to the bottom plate provided a constant heat current to the fluid layer .
the power input was obtained by measuring the voltage across the film heater and the current through it .
the latter was obtained from the voltage across a temperature - controlled standard resistor .
all electrical wires were fed into the rotating frame through an electrical slip ring and inside a hollowed steel shaft which also served as the drive train for rotation .
the electrical noise of the slip ring was small enough that there was no measurable difference in the signals with or without rotation .
rotation was provided by a micro - stepping motor through a shaft , two gears and a timing belt and was under computer control .
the gear ratio set a lower limit for the frequency of about 0.01 hz . the maximum frequency surveyed was 0.5 hz . the convection cell was constructed with aluminum top ( 1.27 cm thick ) and bottom ( 0.64 cm thick ) plates and plexiglass sidewalls ( 0.32 cm thick ) .
the aluminum plates were anodized to prevent corrosion in water .
the cell had a height @xmath30 cm and horizontal dimensions @xmath31 cm with an aspect ratio of @xmath32 .
the cell s bottom and sides were insulated by 2.5-cm - thick styrofoam to reduce thermal losses from radiation and conduction or convection by air .
four thermistors were embedded in each of the top and bottom plates and the thermistor centers were located at 0.32 and 0.23 cm from the fluid , respectively , and at the mid point between the cell s center and the sides .
the four thermistors in each plate gave the average plate temperature .
the wires from the film heater and from the four thermistors were heat sunk on the bottom of the cylindrical can which was maintained at the same temperature as the top plate .
heat is transported more efficiently by convection where heat can be advected by the fluid motion than by conduction where heat is transported solely by diffusion .
the enhancement of thermal transport by convection is characterized by the nusselt number nu @xmath33 where @xmath34 and @xmath35 are the effective thermal conductance and molecular thermal conductance of the fluid layer , respectively . in an experimental realization of convection
there are additional heat transport contributions to the measured thermal conductance .
these include conduction through the cell sidewalls and from conduction , convection , and radiation to the surrounding environment .
below the onset of convection , the background conductivity can be measured and effectively subtracted from the total heat transport contribution provided that the thermal conductivity of water , available from the literature , is assumed .
for turbulent convection in room temperature experiments this is difficult to accomplish because if @xmath36 corresponds to the maximum ra @xmath37 then the onset of convection occurs for @xmath38 , far below experimental resolution
. this can be overcome by rotating the cell , thereby suppressing the onset of convection by about four orders of magnitude , _
i.e. _ , @xmath39 , and allowing the background thermal conductivity to be measured directly .
we now describe this procedure in detail .
the background thermal conductance was determined to be @xmath40 w / k , comparable to the water layer s thermal conductance @xmath41 w / k at a cell mean temperature of 21.5 @xmath42c . assuming that @xmath43 is independent of the mean temperature of the cell , the nusselt number is given by @xmath44 where @xmath45 is the temperature difference across the water layer with @xmath46 and @xmath47 being the top - plate and bottom - plate temperatures
, @xmath48 is the total heat input to the bottom plate , and @xmath35 is the thermal conductance of the water layer at the mean temperature @xmath49 .
the top and bottom temperatures are corrected for the temperature drop in the aluminum plates although this correction was always less than 0.3% . in our experiment , we measured @xmath43 at a single temperature because the temperature dependence of the background terms is quite small and does not affect the data presented here . to evaluate the systematic error in our measurements of @xmath43 , we have estimated the different contributions in that quantity .
the major contributors to @xmath43 are the plexiglass side walls , the insulating foam , the electrical wires , and thermal radiation .
the first three contributions have very weak temperature dependence ( less than 1% change over a 20 k temperature difference ) and small magnitude , estimated to be 0.006 , 0.006 , and 0.003 w / k , respectively .
the rest of the measured background heat transport , about 60% of @xmath43 ( @xmath50 w / k ) , comes from a series combination of conduction through the foam , convection in the air surrounding the insulation , and from thermal radiation from the outer surfaces of the insulation to the surroundings which are maintained at the top - plate temperature .
the top - plate temperature was held constant and the bottom - plate temperature was changed by a maximum of 20 k corresponding to the maximum heat input .
this produces about a 2 k increase in the average temperature of the radiating surfaces , or about a 2% increase in radiated heat , which results in about a 0.02 overestimate in nu .
therefore , for the measurements with fixed @xmath46 , @xmath43 can be taken to be constant as the correction to nu is less than 0.1% .
this analysis neglects an important point regarding the heat transported through the side walls @xcite . rather than supporting a linear temperature profile as in the non - convecting state where the background is measured ,
the side walls are in contact with a turbulent fluid that is approximately iso - thermal in the bulk of the flow .
this leads to an enhancement of heat transport through the side walls in the turbulent state . for conditions similar to those presented here ( thin plexiglass walls relative to the lateral extent of the system ) , however , the correction to the total heat transport was shown by numerical modeling to be small , ranging from about 2% for nu = 10 to 1% at nu = 100 .
our sidewall is thicker by about a factor of two , so in the worst case these values might be twice as large . applying a correction of
this order shifts the exponent of a power law fit by at most 1% ( higher ) and the constant term by about 7% ( lower ) .
these estimates contribute to the systematic error in our results but we do not explicitly correct for the side wall effect in the data presented below .
the parameter space for rotating convection is defined by ra and ta ( or @xmath51 ) which are proportional to the physical control variables of @xmath13 and @xmath6 , respectively . in fig .
[ fig : ra - ta - map ] , the parameter space is shown over a range which encompasses our experimental measurements .
the shaded region denotes the area in which most of our efforts were focused .
the limitations which determine that area are the range of measurable @xmath13 for a given cell height d ; a four - decade variation of ra was obtained by varying the temperature difference across the cell from 2 mk to 20 k. the taylor number for the shaded area ranged from about @xmath52 to @xmath53 , corresponding to rotation frequencies from @xmath54 hz to @xmath55 hz .
the lower limit was determined by the range of stability of the stepping motor given a particular gear ratio . by reducing that ratio ,
lower taylor numbers could have been investigated but this proved unnecessary as the interesting range of ta was spanned with the chosen gear ratio . during the experiments , we fixed some parameters and studied the dependence of measured quantities on the others .
for instance , we fixed ra to study the dependence on ta , and visa versa .
one important parameter , the convective rossby number ro which provides a relative measure of buoyancy relative to rotation , was maintained fixed by varying both @xmath13 and @xmath6 for each data point .
the contours of constant ro , plotted in fig . [ fig : ra - ta - map ] , are approximately straight lines in the log - log plot . numerical simulations of turbulent rotating convection with @xmath56 @xcite followed the contour of ro = 3/4 where buoyancy and rotation had roughly comparable importance . this particular line is noted in the figure .
several other sets of data shown in the figure are discussed later .
overall , the parameter ranges in our experiments fall roughly into the region studied in open - top experiments @xcite ; the parameter space of experiments on rotating convection prior to about 1990 was summarized in @xcite .
( the _ flux _ rayleigh number used in that parameter space is related to ra by ra@xmath57 = nura so that the highest ra in our experiments , where nu @xmath58 corresponds to ra@xmath59 ) .
in this section , we present experimental results for non - rotating convection .
this enables us to compare our results with existing theories and with other non - rotating convection experiments , of which there are many .
it also serves as a reference for our results on rotating convection .
we concentrate here on measurements of heat transport in fluids with prandtl number @xmath60 .
heat transport is measured by the nusselt number nu and scales with ra as a power - law : nu @xmath61ra@xmath22 .
classical arguments @xcite suggest @xmath62 whereas other scaling theories @xcite predict @xmath24 .
another approach that generalizes the latter theories @xcite predicts a complicated phase diagram as a function of ra and @xmath0 . in different regions
different power - law scalings dominate with strong crossover effects so that one obtains a form for the heat transport where nu scales with ra as @xmath63 with @xmath64 and @xmath65 fixed .
careful heat transport measurements have demonstrated that this latter prediction yields better fits to experimental data @xcite . for the limited range of ra presented here , either approach yields a good fit to the data . the nusselt number of non - rotating convection measured with fixed @xmath46 is shown in fig .
[ fig : nu - omega-0 ] as a function of ra for @xmath66 ra @xmath3 .
the results are reasonably well decribed by a power - law with scaling coefficients @xmath67 and @xmath68 which were obtained by fitting all data for @xmath66 ra @xmath3 shown in the figure .
fitting the data between @xmath69 and @xmath70 yields slightly different values : @xmath71 and @xmath72 .
these latter values have the virtue of being derived from a range of ra which is in the turbulence regime but with the disadvantage of a shorter scaling range .
to account for this systematic uncertainty we take @xmath73 and @xmath74 as the best estimates for the scaling coefficients .
the exponent is very close to 2/7 and agrees well with earlier work as summarized in table [ tab : coef ] . although the 2/7 value has been shown not to describe heat transport data over a much larger range of ra @xcite , we use it here for convenience .
further , the uncertainties associated with the exponents and coefficients are the result of statistical fits and almost certainly underestimate the systematic errors associated with the different experiments .
@lllllll@ + @xmath75 & @xmath76 & @xmath77 & @xmath77 & ra range & @xmath20 & reference + & & ( @xmath7 ) & ( @xmath78 ) & ( @xmath79 ) & & + + 0.131 & 0.300(5 ) & 0.165 & 0.170 & @xmath80 & 2.8 - 10 ( c ) & rossby ( 1969 ) + 0.183 & 0.278 & 0.162 & 0.159 & @xmath81 & 1.5 @xmath82 2.5(r ) & chu & goldstein ( 1973 ) + 0.145 & 0.29 & 0.155 & 0.156 & @xmath83 & 3.5 - 14 ( s ) & tanaka & miyata ( 1980 ) + 0.137 & 0.275(7 ) & 0.115 & 0.112 & @xmath84 & 0.71 , 1.6 ( s ) & solomon & gollub ( 1991 ) + 0.129 & 0.299(3 ) & 0.160 & 0.164 & @xmath85 & 2.0 ( c ) & zhong _ et al . _
( 1993 ) + 0.19(4 ) & 0.28(1 ) & 0.172 & 0.170 & @xmath86 & 4 @xmath82 1 ( r ) & chill _ et al . _
( 1993 ) + 0.145 & 0.292 & 0.160 & 0.162 & @xmath87 & 1.0 ( c ) & cioni _ et al . _
( 1996 ) + 0.16 & 0.281(2 ) & 0.147 & 0.145 & @xmath88 & 1.0 ( c ) & shen _ et al . _
( 1996 ) + 0.164(6 ) & 0.286(3 ) & 0.164 & 0.164 & @xmath89 & 0.78 ( s ) & this work + 0.154 & 0.291 & 0.168 & 0.170 & 3000 - 60000 & 1.0 ( c ) & nikolaenko and ahlers ( 2003 ) .
compared to the scaling exponent @xmath76 , the coefficient @xmath75 is quite different from one experiment to another .
@xmath75 is sensitive to the exponent and a precise determination of @xmath75 requires a larger range of ra than has been available in any of the experiments using water .
fixing the exponent at 2/7 and computing a value for @xmath77 ( equivalent to nu / ra@xmath90 ) at different ra gives a better comparison between data sets , see table [ tab : coef ] .
for the water experiments and for @xmath91 , the coefficients agree quite well except for the experiments of solomon and gollub @xcite where a liquid mercury bottom surface may account for the discrepancy .
all of the data reported earlier and listed in the table do not directly measure the background heat transport contribution that we are able to account for using rotation .
this background measurement is important in eliminating systematic error for smaller nu .
an average over all the data sets for convection in water yields @xmath92 and @xmath93 with no statistically - significant dependence of nu on @xmath20 . in summary ,
our data for non - rotating convection agree well with earlier results despite the significant variation in aspect ratios between experiments .
we also took heat transport data at fixed mean - cell temperature @xmath94 to estimate the prandtl number dependence of the heat transport @xcite ( not reported here ) .
we can use that data to correct the data at fixed @xmath46 . in the fixed @xmath46 measurements reported in the remainder of this paper
, we had a variation in the range 21.5@xmath42c @xmath95 31.4@xmath42c ( and resultant @xmath96 ) corresponding to changes from the lowest to highest heat input .
such non - constant @xmath0 or @xmath94 results in an uncertainty in nu of the order of 0.8% .
we interpolate the nu data to a constant mean temperature or a constant @xmath0 .
we choose @xmath97c ( where @xmath98 ) as the reference temperature which was about the average of the mean temperatures in the experiments .
the interpolated value is given by nu = nu@xmath99 where nu@xmath100 is the measured nusselt number using eq .
[ eq : n ] .
the difference between the interpolated and measured values of nu is less than 0.8% , and does not change the scaling coefficients within their specified error bars .
nevertheless , the interpolated values are reported in fig .
[ fig : nu - omega-0 ] .
one important feature of turbulent convection is the large scale circulation that is driven by an accumulation of thermal plumes that congregate near the lateral boundaries @xcite .
the general circulation in our cell was visualized using glass encapsulated thermochromic liquid crystals ( tlc ) . a white light sheet about 1 mm in width was used to illuminate the cell from the side and a black background was provided for good contrast .
the mean - flow direction was typically across the cell diagonal .
flow reversals were observed as was a shifting of the main diagonal circulation in a clockwise or counterclockwise direction as viewed from above .
this meant that sometimes the flow along the off - diagonal direction reversed directions frequently .
this cross diagonal flow has been observed before in convection cells with square cross section @xcite .
in addition to the large diagonal flows there were often small recirculating cells in the corners and along the bottom - side boundary .
viewed from above the thermal plumes near the bottom boundary layer were arranged into coherent sheets which were swept up by the mean flow .
this was also seen in a number of convection experiments in water where the flow is easy to visualize @xcite .
heat transport measurements in the presence of rotation are complicated by the changing influences of buoyancy , proportional to ra , and rotation , proportional to @xmath6 .
the simplest thing to do experimentaly is to fix @xmath6 and vary @xmath13 . because the mean temperature changes with @xmath13 ( fixed @xmath46 )
, the dimensionless @xmath51 changes owing to the temperature - dependent viscosity of water
. this can be as large as 25% over the ra range that we studied . as noted above @xmath0 also changes somewhat but that influence is small .
even constant @xmath51 is not the appropriate variable to hold constant if one wants to evaluate the behavior of nu as a function of ra for a constant rotational forcing . from previous heat transport measurements
@xcite , it appeared that nu at fixed @xmath6 was enhanced by rotation for intermediate ra but seemed to asymptote to the non - rotating value of nu at higher ra .
a convenient measure of rotational forcing @xcite is defined by the `` convective '' rossby number , ro@xmath101 , which is the ratio of a rotational period to a buoyancy time .
so ro @xmath102 should mark the border between strongly rotating convection with ro @xmath103 and weakly rotating convection with ro @xmath104 . in those numerical simulations of rotating convection at fixed ro = 0.75 ,
nu scaled approximately as ra@xmath90 . here
we present heat transport measurements to higher ra than previously @xcite and consider nu as a function of ra at fixed ro .
in addition , we compare our results to earlier ones by rossby @xcite who compiled constant nu contours as a function of ra and ta . to get a good understanding of the whole system , it is useful to consider different slices of the parameter space .
we present them in the order of contours of nu , fixed @xmath6 , fixed ra , and finally fixed ro .
heat transport measurements in rotating convection can be summarized by a contour plot of nu presented in fig .
[ fig : iso - nu ] .
the points on each constant - nu line were obtained by interpolating nu data measured under different controlled conditions ( namely , fixed @xmath6 , ra , and ro ) .
the individual data points are within 1% of the smooth curves for large nu and within 3% for the smallest nu .
figure [ fig : iso - nu ] complements the nu contour plot in fig .
11 of rossby @xcite where the highest nu was 12 , and the combination of the two gives a rather complete description of nu in the parameter space of ra@xmath105 and @xmath106 . as shown in fig .
11 of @xcite for the lower ra and ta range , there is a minimum ra for constant nu or alternately there is a maximum nu at constant ra . at fixed ta , however , nu is a monotonically increasing function of ra . in the following ,
results are presented which elucidate the origin of the maximum nu at fixed ra ( or the minimum ra at fixed nu ) and investigate the variation of nu as a function of ra , ta , and ro . before proceeding with these details ,
however , we can already see the overall trend of heat transport at fixed ra . for low rotation
, nu is rather insensitive to changes in ta ( note the discontinuity in the horizontal axis in fig . [
fig : iso - nu ] ) . in the intermediate range between lines _
l2 _ and _ l1 _ , nu increases with increasing ta , as rotation enhances heat transport . for high enough ta , however , rotation suppresses convection and nu decreases as the onset of bulk convection is approached at ra@xmath107(ta ) . shown in fig .
[ fig : nu - omega-500 ] is nu versus ra at @xmath108 hz where we obtained the background conductance @xmath43 as described in sec .
[ sec : setup ] .
the first few points have larger uncertainty because of the small temperature difference @xmath13 across the cell and the long thermal diffusion time in our cell ( about 16 h ) .
the transition to convection from a conduction state occurred at @xmath109 mk and ra@xmath110 , much lower than the theoretical value of @xmath111 for a laterally - infinite system at this rotation rate @xcite .
this lower - than - expected transition was observed in early heat - transport experiments @xcite and was later visually identified as a transition to a side - wall traveling - wave state @xcite . extrapolating the results in @xcite to @xmath112 ,
one obtains the onset to the traveling state at about @xmath113 , not far from our value of @xmath114 .
there is also an inflection point in nu at ra@xmath115 , coinciding with the theoretical prediction @xcite for the transition to bulk convection @xcite .
note that as the mean temperature of the cell increased during the measurements ( constant @xmath46 ) , @xmath51 and ta increased from @xmath116 and @xmath117 at small ra to @xmath118 and @xmath119 at the highest ra , respectively .
we also measured nu at @xmath120 , 0.188 , 0.502 , and 1.26 hz . these measurements , shown in fig .
[ fig : nu - fix - omega ] , served as a rough characterization of the system .
the first observation is that higher rotation suppresses convection relative to non - rotating convection from onset up to a value of ra that depends on rotation . above that ra ,
nu is higher than its corresponding value without rotation .
thus , the notion that rotation is a damping influence on convection , as suggested by the taylor - proudman theorem , is only valid near onset and the opposite is true for turbulent convection .
the values of ra@xmath121 for this crossover are plotted in fig .
[ fig : ra - ta - map ] as solid circles .
it is the crossover that gives rise to the maxima in the contours of nu .
the second thing to notice is that although nu(r,@xmath51)/nu(ra,0 ) @xmath122 at intermediate ra , it appears to asymptote to 1 at higher ra .
this suggests that buoyancy wins out over rotation at high ra and fixed @xmath51 .
revisiting the presence of mean flow , a feature of non - rotating convection over a large range of ra , we consider mean flow for rotating convection . on the one hand ,
flow visualization in the rotating frame for a cylindrical cell with @xmath123 @xcite used both thermochromic liquid crystal ( tlc ) and particle image velocimetry to determine the flow structure near the upper boundary layer @xcite ) .
the sheet - like plumes evolved under rotation into vortices and for small enough ro , _
i.e. _ , for rotation dominated flow , there was no indication of a large scale circulation extending over the size of the container .
it seems that the shear on the boundary layer is of a very different form than for non - rotating convection as strong vortical motions dominate the flow just outside the boundary layer .
results for higher ra in the range @xmath8 to @xmath124 @xcite indicated a precessing mean flow provided ro @xmath125 . similarly , recent results showed a breakdown of large - scale circulation for ro @xmath126 @xcite .
another way to look at the influence of rotation on convection is to fix ra and vary @xmath51 .
we measured nu as a function of ta at rayleigh numbers of ra @xmath127 , with results as shown in fig .
[ fig : nu - ta - fixra ] . for each ra ,
the data were corrected so that @xmath0 was constant .
we have also included data for ra @xmath128 which were obtained by interpolating data measured at constant ro and @xmath6 .
for all the data sets , nu increased with rotation before decreasing at higher ta with a maximum nu that varied with ra .
the loci ta@xmath129 of maximum nu are plotted in fig .
[ fig : ra - ta - map ] as solid squares and in fig .
[ fig : iso - nu ] ( line @xmath130 ) .
note in the latter figure that this line approximately connects the loci of minimum ra for each nu contour .
line @xmath130 can be approximated by ra @xmath131 , differing from the relation of ra @xmath132 at lower ra and ta reported by rossby ( 1969 ) .
the very different exponents indicate a continuously steepening curve and suggests that there is no clear asymptotic ( in ta ) power - law scaling for the nu maxima over the ta range studied so far .
an interesting conjecture regarding the enhancement of heat transport by rotation is that rotation creates thermal vortices which increase nu through ekman pumping in the boundary layer @xcite .
thus , the enhancement in nu might be proportional to the number of such vortices . because we have not visualized the flow for the rotating system , the number of vortices as a function of ta is not known directly @xcite .
instead we consider the linear prediction for the number of structures at the convective onset .
the linear wavenumber @xmath133 increases with ta and asymptotically scales like ta@xmath134 @xcite which implies that the number of cellular structures should scale like @xmath135 .
this scaling for vortex number was observed even significantly above onset in experiments with an open top surface @xcite which suggests it is a reasonable assumption here . instead of the 1/3 scaling of the asymptotic theory
, however , we will compare with an empirical fit to the linear data over our range of ta which gives @xmath136 . in fig .
[ fig : nu - ta - fixra - fit ] , we plot nu in fig .
[ fig : nu - ta - fixra ] as a function of ta@xmath137 . for ta
@xmath138 ta@xmath100 , there is a linear region for each ra that shrinks as ra decreases .
in the linear region , we have @xmath139 where nu@xmath140 and @xmath141 are fitting parameters and values for different ra are listed in table [ tab : tab1 ] . .fitting parameters nu@xmath140 and @xmath141 .
nu@xmath100 and ta@xmath129 are the maximum nusselt number and its location at constant ra .
nu@xmath142 is the non - rotating value obtained by cubic least - square interpolation of @xmath143(nu ) vs @xmath143(ra ) for non - rotating convection . [
cols="^,^,^,^,^,^,^ " , ] solid lines in figs . [
fig : nu - ta - fixra ] and [ fig : nu - ta - fixra - fit ] are calculated from these fitting parameters .
the deviation from the solid lines ( linear behavior ) at higher ta is a result of rotation suppressing convection in the weakly nonlinear regime near onset .
also listed in table [ tab : tab1 ] is nu@xmath144 , the nusselt number of non - rotating convection .
the fitting parameter nu@xmath140 for all ra is nearly identical to nu@xmath144 within fitting and experimental uncertainties .
this indicates that the fitting is consistent with the data in the range 0 @xmath145 ta @xmath145 ta@xmath129 .
thus , the enhancement of nu by rotation is given by @xmath146nu = @xmath141ta@xmath147 with @xmath141 about 0.006 .
this result supports the conjecture that the enhancement is proportional to the average number of thermal vortices . using velocity field measurements @xcite
, the number of vortices could be evaluated more quantitatively . at a fixed ra = @xmath148 ,
ta was varied from 0 up to about @xmath149 .
the variation of vortex ( cyclonic ) density with @xmath51 was slightly sub - linear , over the range @xmath150 .
since @xmath151 , this result suggests that the number of vortices depends on ta with a power a bit less than 1/2 .
this is roughly consistent with the estimate based on linear stability arguments .
unfortunately , the vortex density data are too sparse to provide better estimates in the ta range of interest .
many experiments in thermal convection without rotation showed that nu scales more closely as the 2/7 power of ra in the regime @xmath152 than with the classical 1/3 power law .
as discussed earlier , a generalized theory in terms of a phase diagram in ra and @xmath0 and more precise experimental measurements suggest a form with the sum of two power laws with exponents of 1/5 , 1/4 , 1/3 , or 1/2 depending on the region in phase space .
rotation complicates the issue of scaling since the relative influence of rotation changes with changing ra at fixed ta .
numerical simulations @xcite showed that the convective rossby number is a good measure of the relative importance of buoyancy with respect to rotation : for ro = 0.75 and @xmath153 , nu @xmath154 ra@xmath90 which indicates that the details of rotation are relatively unimportant in the determination of the scaling behavior .
we have tested this prediction and over the range @xmath155 find approximate 2/7 power - law scaling .
we use this single power law description for convenience - a fit of the form nu = a ra@xmath156 + b ra@xmath157 yields equivalent fits . for the non - rotating case ,
the coefficients are a = 0.26 and b=0.047 we have measured nu at several constant rossby numbers , i.e. , ro = 0.12 , 0.30 , 0.52 , 0.75 , 1.15 , 1.49 , and @xmath158 ( zero rotation ) . for clarity , only part of the data for ro = 0.30 , 0.75 , and infinity are plotted on a log - log scale in fig .
[ fig : nu - fix - ro ] with the coefficients of power law fits listed in the plot .
the nusselt number agrees well with a 2/7 power law and is different from the 1/3 power law , especially at higher rayleigh number ( @xmath159 ) . in fig .
[ fig : nu - fix - ro - scaled ] , we plot nu / ra@xmath90 versus ra to gauge how well the 2/7 power law describes the data . over the ra range where the 2/7 power law is satisfied ,
a constant value of @xmath160 is expected .
the description is reasonably good for the non - rotating case , but becomes a little worse as ro decreases or rotation increases .
the coefficient @xmath77 is plotted in fig . [
fig : a - a - ro](c ) .
the trend of increasing @xmath77 with decreasing ro demonstrates the enhancement of heat transport by rotation .
we also fit the data with nu @xmath161 to obtain the coefficient and the exponent as functions of ro .
least - squares fitting was performed in the range @xmath162 to avoid possible deviation from power - law scaling at lower ra .
the results are plotted in fig .
[ fig : a - a - ro ] .
it should be pointed out that for non - rotating convection the coefficient @xmath163 and exponent @xmath164 are slightly different from the values obtained by fitting the data for @xmath165 . for finite ro ,
the exponent depends on ro : it decreases almost monotonically from 0.287 at ro = 1.5 to 0.269 at ro = 0.12 . as shown in figs .
[ fig : nu - fix - ro ] , [ fig : nu - fix - ro - scaled ] , and [ fig : a - a - ro ] , the coefficient @xmath75 increases as rotation increases , and its value is very sensitive to the fitting value of exponent as can be seen by comparing fig .
[ fig : a - a - ro](b ) and ( c ) .
determining unambiguously the scaling behavior of the heat transport requires many orders of magnitude in ra .
thus , an absolute comparison of scaling exponents in our experiment is uncertain . nonetheless , our experiments yielded some interesting results , especially when compared to numerical simulations @xcite .
first , at fixed ro , nu depends on ra with a power law close to 2/7 , in agreement with numerical simulation @xcite for ro = 0.75 and @xmath56 .
if anything , rotation seems to reduce the scaling exponent slightly .
this could be the result of different scaling ranges as a function of ro , because fixed ro does not exactly maintain a balance between buoyancy and rotation , or because rotation modifies the scaling exponent directly .
an extended range in ra would be necessary to resolve this quantitatively , perhaps in a gas system . from the perspective of turbulent convection theory @xcite ,
the insensitivity of the scaling to rotation is rather interesting because rotation affects many properties of the turbulence , such as the change from thermal plumes to vortices and the existence of a turbulent eckman boundary layer and associated eckman pumping . in non - rotating convection ,
the relationship between the thermal boundary layer thickness @xmath166 and the viscous sublayer thickness @xmath167 determines the power - law scaling in the sheared boundary layer theory @xcite where nu @xmath154 ra@xmath90 applies when @xmath168 .
rotation introduces another vertical length scale , the ekman layer thickness @xmath169 , which could in principle play a role similar to @xmath167 in non - rotating convection . in fig .
[ fig : bl - scaled ] , we plot @xmath166 and @xmath169 as functions of ra at ro=0.30 , 0.75 , 1.49 where @xmath166 and @xmath169 are defined as @xmath170 numerical simulations @xcite showed that the transition to turbulent scaling occurs at ra @xmath171 for ro=0.75 where @xmath172 .
in our experiment we also have @xmath172 at ro = 0.75 despite the different prandtl number used in the experiment .
thus , the condition @xmath172 for a turbulence transition is not consistent with our experiments .
it is also interesting to notice that at high ra @xmath173 and constant ro , @xmath174 decreases faster than @xmath175 ra@xmath176 , therefore @xmath166 and @xmath169 cross at high ra for ro @xmath145 0.75 and do not cross at ra for ro @xmath177 0.75 .
the interplay of these two length scales is as yet not understood nor is it clear that @xmath169 defined from non - convecting problems with differential rotation is the proper variable to use here .
it is perhaps interesting to note that along lines of constant ro , the ratio @xmath178 is almost constant .
a direct measurement of the turbulent ekman layer in the presence of a thermal boundary layer would be very useful to augment the arguments based on numerical simulations @xcite for a coexisting thermal boundary layer with a linear ekman layer .
subsequent flow visualization @xcite using particle image velocimetry showed an diverse set of interesting behaviors of velocity and vorticity fluctuations but could not yield a definitive conclusion regarding the complex interplay of thermal and kinetic boundary layers involved in determining heat transport for turbulent convection .
we have presented experimental studies of turbulent thermal convection in water confined in a cell with a square cross section with and without rotation . in non - rotating convection , the nusselt number was found to scale roughly as ra@xmath90 at @xmath179 .
heat transport measurement in rotating convection confirmed the findings by other researchers @xcite that rotation enhances thermal transport over a certain range of ra and ta range .
evidence showed that such enhancement may be attributed to the increased effective horizontal area caused by the presence of vortices under rotation . at fixed rossby number
, nu was found to scale approximately as ra@xmath90 , as does nu of non - rotating convection and predicted from numerical simulation @xcite .
characterization by a combination of power laws @xcite was equally good at fitting the data for both rotating and non - rotating convection .
analysis using ekman layers instead of kinetic boundary layers as input into a scaling theory did not provide additional insight into the heat transport data , and it remains unclear how rotation and its associated modification of boundary layer structure affects heat transport .
given the extensive study of non - rotating convection in recent years with great advances in characterizing boundary layers , heat transport and large scale circulation @xcite , there seems to be an emerging opportunity to apply similar rigor to the geophysically important case of rotating thermal convection .
we hope that our work may stimulate more experimental and theoretical studies on rotating thermal turbulence . a rigorous test of the power - law scaling of nu under constant ro and the validity of ro as the good parameter requires a much larger ra and ta range than what was available in our experiment .
a gas convection may be needed to achieve such experimental conditions .
the interplay amongst various length scales , such as horizontal vortex wavelength , ekman layer , thermal boundary layer , viscous sublayer , was not well understood in our experiment and certainly calls for further work on this aspect of turbulent convection . | we present experimental heat transport measurements of turbulent rayleigh - bnard convection with rotation about a vertical axis .
the fluid , water with prandtl number ( @xmath0 ) about 6 , was confined in a cell which had a square cross section of 7.3 @xmath17.3 cm and a height of 9.4 cm .
heat transport was measured for rayleigh numbers @xmath2 ra @xmath3 and taylor numbers @xmath4 ta @xmath5 .
we show the variation of normalized heat transport , the nusselt number , at fixed dimensional rotation rate @xmath6 , at fixed ra varying ta , at fixed ta varying ra , and at fixed rossby number ro .
the scaling of heat transport in the range @xmath7 to about @xmath8 is roughly 0.29 with a ro dependent coefficient or equivalently is also well fit by a combination of power laws of the form @xmath9 .
the range of ra is not sufficient to differentiate single power law or combined power law scaling .
the overall impact of rotation on heat transport in turbulent convection is assessed . |
if @xmath2 is a compact complex manifold , the algebraic dimension of @xmath2 , usually denoted by @xmath3 , is defined to be the complex dimension of a projective algebraic variety , the rational function field of which is isomorphic to that of @xmath2 .
we always have @xmath4 , and in case of equality , @xmath2 is called moishezon .
any projective algebraic manifold is moishezon . in the other extreme case ,
@xmath5 , @xmath2 has no non - constant rational function .
not so many compact differential manifolds admit complex structures whose algebraic dimension ranges from zero to half of its real dimension .
complex tori of dimension @xmath6 are examples of such manifolds . in complex dimension two ,
only complex tori and k3 surfaces are examples for which all three possible values of the algebraic dimension actually occur . in dimension three , _ twistor spaces _ associated to self - dual metrics on 4-manifolds @xcite are good candidates of such manifolds . by a result of campana @xcite , if @xmath2 is a moishezon twistor space , then the base 4-manifold is homeomorphic to @xmath0 , the connected sum of @xmath7 copies of complex projective planes ; by convention @xmath8 .
if @xmath9 , any twistor space on @xmath0 is moishezon as long as the corresponding self - dual metric has positive scalar curvature @xcite .
no example seems to be known of a self - dual metric of non - positive scalar curvature on these manifolds . thus if @xmath9 , all known twistor spaces are moishezon .
the situation is very different for @xmath10 .
first , if @xmath11 , for any @xmath12 , there actually exists a twistor space @xmath2 over @xmath13 which satisfies @xmath14 , see @xcite .
also , it is known that @xmath15 as long as the self - dual metric on @xmath13 is of positive scalar curvature @xcite .
moreover , no example seems to be known of a self - dual metric on @xmath13 of non - positive scalar curvature .
thus our understanding of possible values of the algebraic dimension is quite satisfactory in case @xmath11 , too .
the main focus of this article is on the case @xmath1 . for any @xmath1 and @xmath16
it is known that there exists a twistor space @xmath2 on @xmath0 which satisfies @xmath14 , @xcite . moreover , in ( * ? ? ?
* main theorem ) , it is stated that there exists a twistor space @xmath2 on @xmath0 with @xmath17 for any @xmath1 .
these spaces satisfy @xmath18 , where @xmath19 is the natural square root of the anti - canonical line bundle on @xmath2 , which is available on any twistor space @xcite .
this is in contrast to the following theorem , which is main result of this article .
[ thm ] if @xmath1 and @xmath2 is a twistor space on @xmath0 such that @xmath20 , then @xmath21 . at the end of this article
we give a detailed explanation of the contradiction between this result and the result of @xcite . by a result of @xcite ,
if a twistor space @xmath2 over @xmath0 , @xmath1 , satisfies @xmath22 , then @xmath2 is moishezon .
therefore from theorem [ thm ] , we obtain if a twistor space @xmath2 over @xmath0 , @xmath1 , satisfies @xmath17 , then the system @xmath23 consists of a single member , or is empty . to the best of the authors knowledge , existence of this kind of twistor spaces
is not known .
thus the existence of a twistor space on @xmath0 with algebraic dimension two seems to be not known .
here is an outline of our proof of theorem [ thm ] .
assume @xmath1 and @xmath2 is a twistor space over @xmath0 which satisfies @xmath20 and @xmath17 .
it can be seen that the base locus of the pencil @xmath23 is a cycle of smooth rational curves . by blowing up this base curve and then taking a simultaneous small resolution of all ordinary double points that appear by the blow - up
, we get the diagram @xmath24 here , @xmath25 is the composition of the blow - up and the small resolution , and @xmath26 is a surjective morphism induced by the pencil @xmath27 , where @xmath28 is the exceptional divisor of the birational morphism @xmath25 .
fibres of @xmath26 are naturally identified with members of the pencil @xmath23 .
the exceptional divisor @xmath28 plays a significant role in our proof .
the morphism @xmath26 naturally gives it the structure of a fibre bundle @xmath29 except over a finite number of points .
the typical fibre of is a cycle of rational curves , which is naturally identified with the base curve of the pencil @xmath23 .
fibres over the exceptional points are cycles of rational curves as well , but they have two more irreducible components than the typical fibres , because the exceptional curves of the small resolution are inserted at the ordinary double points of the blow - up .
the presence of these structures follows from the assumption that @xmath30 includes a pencil whose base locus is a cycle of rational curves .
when @xmath17 , any smooth fibre of @xmath26 has to be a rational surface whose anti - kodaira dimension is one . by making use of the zariski decomposition of an anti - canonical curve on the surface ,
we explicitly define a line bundle @xmath31 over @xmath32 which satisfies the following properties : * an algebraic reduction of @xmath32 is realised by the linear system @xmath33 for @xmath34 , * @xmath31 is trivial along a typical fibre of the fibration . roughly speaking ,
the line bundle @xmath31 is the ` strict transform ' onto @xmath32 of the line bundle @xmath35 over @xmath2 for some concrete integers @xmath36 and @xmath37 , which are again obtained from the zariski decomposition .
thus the problem of determining @xmath3 is reduced to computing the dimension @xmath38 for arbitrary @xmath39 .
the first observation for this computation is that the line bundle @xmath31 has negative degree along a component of some exceptional fibres of the fibration . from this , together with some properties of the zariski decomposition for the anti - canonical cycle , we can deduce that , for any @xmath40 , the movable part of the linear system @xmath33 is composed with the pencil @xmath41 , and this gives the desired conclusion . in the actual computation ,
we introduce an auxiliary line bundle @xmath42 over @xmath32 for @xmath43 , which coincides with @xmath44 when @xmath45 takes some special value , and show that the system @xmath46 is always composed with the pencil @xmath47 .
finally , a remark about notation . if @xmath48 is a complete linear system on some compact complex manifold and if we write @xmath49 for some effective divisor @xmath50 , then @xmath50 is a fixed component of @xmath48 .
but this does not imply that the system @xmath51 is without fixed component .
we would like to express our sincere gratitude to professor fujiki for his generous and helpful comments about the anti - kodaira dimension of general members of the pencil @xmath23 on the twistor space .
these are reflected not only in the explanation in section 4 but throughout this article .
in this section we fist recall basic properties of the zariski decomposition of a divisor on a projective surface , and then investigate the zariski decomposition of an anti - canonical divisor on a non - singular rational surface of anti - kodaira dimension zero or one .
let @xmath52 be a non - singular projective surface , and @xmath53 an effective divisor on @xmath52 .
then the _ zariski decomposition _ of @xmath53 is the decomposition @xmath54 where @xmath55 and @xmath56 are effective @xmath57-divisors ( i.e. divisors with positive rational coefficients ) or the zero divisor , which satisfy 1 .
@xmath55 is nef , that is @xmath58 for any curve @xmath50 on @xmath52 , 2 . if @xmath59 , @xmath56 is negative definite in the following sense : if @xmath60 with distinct , irreducible @xmath61 and @xmath62 , the total intersection matrix @xmath63 is negative definite , 3 .
if @xmath59 , @xmath64 for all @xmath65 .
any effective divisor @xmath53 admits a unique zariski decomposition , see @xcite .
@xmath55 is called the _ nef part _ of @xmath53 .
note that if @xmath66 is the zariski decomposition of @xmath53 , then for any integer @xmath67 , @xmath68 is the zariski decomposition of @xmath69 .
one of the most important properties of the zariski decomposition is that it disposes of sub - divisors of @xmath69 which do not contribute to the dimension @xmath70 in the following sense : [ prop : z1 ] ( ( * ? ? ?
* lemma 2.4 ) ) let @xmath71 and @xmath72 be as above , and @xmath67 an integer .
then we have @xmath73 where for an effective @xmath57-divisor @xmath50 , @xmath74 and @xmath75 denote the ( integral ) round down and the round up of @xmath50 respectively .
in particular , the divisor @xmath76 is a fixed component of @xmath77 .
next let @xmath52 be a non - singular rational surface .
for an integer @xmath67 , the linear system @xmath78 on @xmath52 is called the @xmath79-th anti - canonical system .
if this is non - empty , the associated rational map @xmath80 , @xmath81 , is called the @xmath79-th anti - canonical map . the anti - kodaira dimension , @xmath82 , of a rational surface @xmath52
is defined as @xmath83 here , @xmath84 occurs when @xmath85 for all @xmath67 , but we will not encounter this case in the following .
typical examples of rational surfaces with @xmath86 are del pezzo surfaces , while simple examples with @xmath87 are obtained from @xmath88 by blowing up the 9 intersection points of two anti - canonical ( i.e. cubic ) curves . the anti - kodaira dimension of rational surfaces was investigated in more detail by f. sakai @xcite .
his main tool was the zariski decomposition of the anti - canonical divisor .
even if @xmath89 , the anti - canonical divisor might not be effective .
however , sakai ( * ? ? ?
* lemma 3.1 ) has shown that the divisor @xmath90 is pseudo effective iff @xmath89 .
a divisor @xmath53 on a surface @xmath52 is called pseudo - effective if @xmath91 for any ample divisor @xmath92 .
fujita @xcite has shown that any pseudo effective divisor @xmath53 has a unique zariski decomposition @xmath66 with the same properties as above except that @xmath55 is no longer required to be effective . therefore ,
if @xmath89 and @xmath93 is the zariski decomposition of an anti - canonical divisor , one can define _ the degree _ of a surface @xmath52 to be the self - intersection number @xmath94 which is clearly a non - negative rational number .
rational surfaces with @xmath95 may be characterised in terms of the degree as follows .
[ prop : degree ] ( sakai ( * ? ? ?
* proposition 1 ) ) let @xmath52 be a non - singular rational surface with @xmath89 , and let @xmath96 be the degree of @xmath52 as above
. then @xmath97 if and only if @xmath98
. in particular we have @xmath99 if ( and only if ) @xmath100 . in the rest of this section we focus on rational surfaces with @xmath99 which admit a special type of anti - canonical divisor . for this purpose , by a _
cycle of rational curves _
, we mean either a rational curve with one node , or a connected , reduced , normal crossing divisor @xmath101 on @xmath52 , with all @xmath102 being non - singular rational curves , such that the dual graph of @xmath53 is a circle . for simplicity , in the expression
, we allow @xmath103 to be one , in which case @xmath104 means a rational curve with one node . we always assume that if @xmath105 in the components @xmath102 and @xmath106 intersect and @xmath107 . by an _
anti - canonical cycle _ we mean a cycle of rational curves which belongs to the anti - canonical class .
the following well - known lemma will be used frequently .
[ lem : negdef ] ( ( * ? ? ?
* , lemma ) ) let @xmath52 be a smooth surface , @xmath108 irreducible curves on @xmath52 for which @xmath109 is a connected curve , @xmath110 rational numbers so that @xmath111 satisfies @xmath112 for @xmath113 .
then , the intersection matrix @xmath114 is negative semi - definite with one - dimensional kernel generated by @xmath55 .
more precisely , this means that , for @xmath115 with @xmath116 we always have @xmath117 and @xmath118 occurs iff @xmath119 for some @xmath120 . for the rest of this section , let @xmath52 be a non - singular rational surface and suppose that @xmath52 has an anti - canonical cycle @xmath53 as in and let @xmath66 be its zariski decomposition .
let @xmath121 be the smallest positive integer for which @xmath122 is integral , and write @xmath123 [ lem : pzero ] 1 . if @xmath124 then @xmath125
2 . @xmath126 if and only if @xmath127 for @xmath128 .
\(i ) if @xmath124 , the nef part of the zariski decomposition for @xmath69 is @xmath129 for any @xmath67 . by proposition [ prop : z1 ]
this means that @xmath130 for all @xmath67 , hence @xmath125 .
\(ii ) from it is clear that @xmath127 for @xmath128 implies @xmath126 .
for the converse , if an index @xmath65 satisfies @xmath131 in , then @xmath56 includes @xmath102 , and so @xmath127 from the property ( iii ) of the zariski decomposition .
therefore we have @xmath132 further , as @xmath55 is nef , we have @xmath133 for any @xmath65 . because we assume @xmath134 ,
implies that @xmath135 even when @xmath136 .
@xmath137 by @xmath138 we denote the group of line bundles on the anti - canonical cycle @xmath53 that are trivial on each component @xmath139 .
thus , part ( ii ) of lemma [ lem : pzero ] says that @xmath126 is equivalent to @xmath140 .
[ lem : blow - down ] let @xmath141 be the blow - down of a @xmath142-curve on @xmath52 , and let @xmath143 be the image of the anti - canonical cycle @xmath53 , and @xmath144 the image of the @xmath142-curve .
define @xmath145 . then , 1 .
@xmath146 is an anti - canonical cycle on @xmath147 and @xmath148 ; 2 . if @xmath149 is the zariski decomposition of @xmath146 and @xmath150 , then @xmath124 iff @xmath151 .
\(i ) let @xmath152 be the exceptional curve of the blow - up @xmath153 .
as @xmath154 , either @xmath28 is a component of @xmath53 or it intersects @xmath53 transversally at one point .
let @xmath155 be the strict transform of @xmath156 , then @xmath157 if @xmath28 is not a component of @xmath53 , or @xmath158 if @xmath28 is a component of @xmath53 . in both cases
, @xmath146 clearly is a cycle of rational curves . because @xmath159 and @xmath160
, we obtain @xmath161 if @xmath28 is not a component of @xmath53 , and @xmath150 if @xmath28 is a component of @xmath53 .
in particular , @xmath162 in both cases . because @xmath163 and @xmath164 , we now obtain @xmath165 .
\(ii ) as @xmath150 , the exceptional curve @xmath28 is a component of @xmath53 .
we fix notation so that @xmath166 and @xmath167 , where @xmath168 .
consider the intersection matrix @xmath169 .
the intersection matrix @xmath170 is obtained from @xmath171 by adding an extra row and a column with @xmath172 and @xmath173 being their only non - zero entries .
in addition , four entries in @xmath171 have to be changed to get @xmath174 , namely @xmath175 , @xmath176 and @xmath177 , whereas @xmath178 . by adding the @xmath103-th column of @xmath174 to columns @xmath179 and @xmath180 ,
the values of the entries of the part of @xmath174 that corresponds to @xmath171 are restored to the original values they had in @xmath171 .
it now follows easily from sylvester s criterion that the matrix @xmath174 is negative definite if and only if @xmath171 is so . because @xmath151 is equivalent to @xmath171 being negative definite and @xmath124 if and only if @xmath174 is negative definite , the claim now follows .
we will frequently need the following more detailed properties of the zariski decomposition @xmath66 of the anti - canonical cycle @xmath53 on the surface @xmath52 .
[ prop : z2 ] suppose that @xmath181 and @xmath126 . 1 .
we have @xmath182 for all @xmath65 .
2 . if @xmath183 is the blow - up of a point @xmath184 and @xmath185 , then @xmath186 3 .
we have @xmath187 for some index @xmath65 .
4 . if @xmath188 , then @xmath189 , @xmath105 , and we have @xmath190 for some indices @xmath65 and @xmath191 . for ( i ) , if @xmath192 for some @xmath65 , we have @xmath193 since @xmath53 is a cycle . but as @xmath126 , lemma [ lem : pzero ] ( ii ) shows that @xmath135 for all @xmath65 , hence @xmath194 . repeating this argument
, we obtain @xmath192 for any @xmath65 , but this contradicts @xmath195 . to prove ( ii ) we recall from ( * ?
* lemma 5.1 ) that @xmath196 with equality if @xmath197 . because @xmath53 is a cycle of rational curves , the points on @xmath53 have multiplicity @xmath179 or @xmath198 .
we assume now @xmath161 and show @xmath199 . from ( i ) , lemma [ lem : pzero ] ( ii ) and lemma [ lem : negdef ] , for any divisor @xmath200 , we have @xmath117 with equality only if @xmath119 for some @xmath120 .
let @xmath201 be the strict transform of @xmath139 and @xmath202 . because we assumed @xmath161
, @xmath203 is an anti - canonical cycle on @xmath204 .
we now show that the intersection matrix of the components of @xmath203 is negative definite . to fix notation , we assume @xmath205 , so that @xmath206 .
consider @xmath207 and let @xmath208 , then @xmath209 . if @xmath210 we then have @xmath118 and @xmath211 , hence @xmath212 for some @xmath213 and so @xmath214 . as @xmath215
this implies @xmath216 , hence @xmath217 , which gives @xmath218 .
this shows that the intersection matrix @xmath219 is negative definite .
this implies that @xmath220 in the zariski decomposition @xmath221of the anti - canonical divisor @xmath203 , hence @xmath222 by lemma [ lem : pzero ] ( i ) .
to prove ( iii ) , we recall from ( * ? ? ?
* theorem 3.4 ) that there is a birational morphism @xmath223 with @xmath224 being non - singular , such that @xmath225 and any anti - canonical divisor @xmath226 is nef , that is @xmath227 in the zariski decomposition @xmath228 of @xmath229 . because @xmath126 and @xmath225 , proposition [ prop : degree ] implies that @xmath230 as well .
the morphism @xmath231 is a composition of blow - ups . by lemma [ lem : blow - down ] ( i ) the image of @xmath53 in each of these partial blow - ups is an anti - canonical cycle containing the blown - up point .
moreover , as we have seen in the proof of ( ii ) above , if the blown - up point had multiplicity one on the anti - canonical cycle , the nef part of the zariski decomposition would vanish after the blow - up . because of lemma [ lem : blow - down ] ( ii ) , this would lead to @xmath124 on @xmath52 , in contradiction to our assumption .
therefore , it follows that at each step a double point of the anti - canonical cycle is blown - up and the nef part of the zariski decomposition does not vanish . because the anti - kodaira dimension does not increase under blow - up , ( *
* lemma 5.1 ) , and @xmath225 , all partial blow - ups in this process have @xmath232 , in particular the self - intersection number of the nef part of the zariski decomposition is equal to zero . from the above we infer that @xmath233 is an anti - canonical cycle on @xmath224 which coincides with the nef part , @xmath234 , of its zariski decomposition .
thus , @xmath235 and all @xmath236 on @xmath224 , and so the assertion ( iii ) holds for the pair @xmath237 .
to finish the proof of ( iii ) by induction , it remains to show that , if @xmath238 is a pair consisting of a rational surface and an anti - canonical cycle on it that satisfies the assumptions of the current proposition and has @xmath236 for one index @xmath65 , and if @xmath183 is the blow - up at a double point of @xmath53 , then the nef part of the new anti - canonical cycle @xmath203 on @xmath204 also satisfies ( iii ) . to show this
, we continue to use the notation introduced above , like , for @xmath239 and @xmath55 . to fix notation , we suppose that the point @xmath240 is blown up by @xmath153 .
let @xmath241 be the exceptional divisor of @xmath153 and write @xmath242 for the strict transform of the component @xmath102 ( @xmath243 ) .
we recall that @xmath244 ( @xmath128 ) , define @xmath245 and consider the divisor @xmath246 on @xmath204 @xmath247 it is easy to see that @xmath248 for @xmath249 , and therefore @xmath246 is nef and @xmath250 . hence ,
again by lemma [ lem : negdef ] , the intersection matrix @xmath251 is negative semi - definite , and if @xmath252 is the nef part of @xmath253 and @xmath254 is the smallest positive integer for which @xmath255 is integral , we have @xmath256 for some @xmath257 . since @xmath187 for some index @xmath65 ( @xmath258 ) by the inductive assumption , it follows that @xmath259 .
if @xmath260 , all the coefficients of @xmath255 would be divisible by @xmath45 . if all @xmath201 appeared in the negative part , @xmath261 , of the zariski decomposition of @xmath203 , condition ( ii ) in the definition of the zariski decomposition would imply @xmath262 .
thus , @xmath263 implies that at least one of the @xmath201 is missing in @xmath261 .
hence , at least one of the coefficients of @xmath255 is equal to @xmath254 .
but then @xmath45 would divide @xmath254 and all coefficients of @xmath255 , which contradicts the choice of @xmath254 .
this shows that @xmath264 , and we obtain @xmath265 .
thus from the expression we now see that the cycle @xmath203 satisfies the property ( iii ) .
the final item ( iv ) follows now easily , because @xmath266 implies that the morphism @xmath231 in the proof of ( iii ) is not an isomorphism and so there was at least one blow - up carried out .
the expression for @xmath122 shows then that at least one of the @xmath267 is greater than @xmath179 . also , note that @xmath268 is equal to the largest coefficient in @xmath122 .
@xmath137 [ prop : z21 ] suppose @xmath181 , @xmath126 and that there exists an integer @xmath39 for which @xmath269 , then @xmath270 has finite order . moreover , if @xmath45 is the smallest positive integer for which @xmath271 , then @xmath272 .
let @xmath45 be the smallest positive integer for which @xmath271 and let @xmath273 be a non - zero element that satisfies @xmath274 .
put @xmath275 and write @xmath276 where @xmath277 ( which may be @xmath129 at this moment ) does not include @xmath102 for any @xmath65 . as @xmath278 , we have @xmath279 for at least one @xmath65 .
since @xmath280 , we have a linear equivalence @xmath281 collecting all indices that satisfy @xmath282 ( if any ) on the left - hand side , we obtain @xmath283 if rhs is the zero - divisor , then so has to be lhs , which contradicts @xmath279 for some @xmath65 .
so both sides of are effective divisors . because lhs and rhs do not have a common irreducible component
, we obtain that the self - intersection number of rhs is non - negative .
since @xmath126 , by lemma [ lem : pzero ] ( ii ) we have @xmath135 for any @xmath65 , @xmath128 . because of proposition [ prop : z2 ] ( i ) we can apply lemma [ lem : negdef ] to obtain that the total intersection matrix @xmath284 is negative semi - definite .
hence the self - intersection number of rhs has to be zero . by lemma [ lem : negdef ]
, this can happen only when @xmath285 for some @xmath286 . because @xmath182 for all @xmath65 by proposition [ prop : z2 ] ( i ) , we can conclude that @xmath287 for all @xmath65 .
therefore from we obtain a linear equivalence @xmath288 on the other hand , we have @xmath187 for some index @xmath65 by proposition [ prop : z2 ] ( iii ) . because @xmath277 is integral , implies now that @xmath289 .
hence by the minimality of @xmath45 , and as @xmath290 , we have @xmath291 . because we have chosen @xmath277 not to have any @xmath139 as a component and because @xmath292 , the divisor @xmath277 does not intersect any @xmath102 . as @xmath293 , this means that @xmath294 .
hence @xmath295 is of finite order in @xmath138 .
@xmath137 we will use this result to show the following property regarding pluri - anti - canonical systems on @xmath52 , which will be used in the next section .
[ prop : z3 ] if @xmath296 , @xmath126 and @xmath140 is of finite order @xmath37 , then 1 .
@xmath297 , @xmath298 , @xmath299 , and the associated morphism @xmath300 is an elliptic fibration ; 2 .
@xmath301 ; 3 . for any @xmath39 ,
the system @xmath302 is composed with the pencil @xmath303 , i.e. each element of @xmath304 is a sum of elements of @xmath305 ; 4 . for any integer @xmath39 , we have @xmath306 because @xmath53 is an anti - canonical divisor with zariski decomposition @xmath66 , we obtain @xmath307 . by lemma [ prop : z21 ]
we have @xmath308 note that @xmath309 is an effective divisor by proposition [ prop : z2 ] ( i ) . using serre - duality
this implies @xmath310 . using @xmath126
, the definition of the zariski decomposition implies that @xmath311 with equality iff @xmath312 .
if @xmath313 , the non - empty system @xmath314 has a ( non - zero ) fixed component . indeed
, if @xmath315 is a line bundle on a smooth surface so that @xmath48 is not empty and @xmath316 , then @xmath48 must have a fixed component , as can be seen as follows .
let @xmath317 with @xmath318 and prime divisors @xmath319 .
then there exists @xmath103 such that @xmath320 , because otherwise we would have @xmath321 .
if now @xmath322 is any element of @xmath48 , then @xmath323 , hence @xmath324 is a component of @xmath322 , i.e. a fixed component of @xmath48 . if @xmath266 we let @xmath50 be the ( maximal ) fixed component of @xmath314 . by lemma [ lem : pzero ] ( ii ) , proposition [ prop : z2 ] ( i ) and lemma [ lem : negdef ] , the intersection matrix formed by the components of the cycle @xmath53 is negative semi - definite , hence @xmath325 .
but if this was negative , @xmath326 would still have a fixed component , in contradiction to the choice of @xmath50 .
hence @xmath327 .
if @xmath328 , we simply take @xmath217 to obtain @xmath327 . therefore , by lemma [ lem : negdef ] again , we have @xmath329 moreover , @xmath330 is an integer since @xmath122 has a component of multiplicity one , by proposition [ prop : z2 ] ( iii ) , and the left - hand side is an integral divisor .
from we have @xmath331 , and so @xmath332 by
. therefore , @xmath333 by riemann - roch we readily obtain @xmath334 . because @xmath335 and @xmath336 , we must have @xmath337 .
since @xmath338 , the standard exact sequence @xmath339 implies now that @xmath340 .
in addition , this sequence provides us with a surjection @xmath341 , from which we obtain @xmath342 .
the general fibre of the morphism @xmath343 associated to the pencil @xmath303 is non - singular , because @xmath52 is smooth .
the smooth ( and hence all ) fibres of @xmath344 are connected . to see this ,
let @xmath345 be a smooth fibre of @xmath344 , then @xmath346 and @xmath347 , hence @xmath348 . because @xmath349 , the exact sequence @xmath350 implies @xmath351 .
therefore , @xmath352 , i.e. @xmath345 is connected .
then , since @xmath353 , the genus formula implies that the general fibre of @xmath344 is an elliptic curve .
this completes the proof of ( i ) . to prove ( ii ) we just have to observe that ( i ) immediately implies @xmath354 .
the assumption @xmath126 , on the other hand , gives @xmath355 , and so @xmath301 . to prove assertion ( iii ) ,
first observe that @xmath356 .
as @xmath344 has connected fibres , we have @xmath357 and the projection formula implies @xmath358 .
hence , @xmath359 , which shows that each element of @xmath304 is a sum of fibres , i.e. a sum of elements of @xmath305 .
finally , for ( iv ) , let @xmath360 be as above and let @xmath361 be the point for which @xmath362 .
let @xmath363 be any member . then since @xmath364 and each member of this linear system is a sum of members of @xmath303 , i.e. fibres of @xmath344 , and because @xmath365 there exist points @xmath366 such that @xmath367 this implies assertion ( iv ) , because @xmath368 , as we have shown above .
@xmath137 [ cor : finord ] suppose that @xmath181 and @xmath126 .
then @xmath369 if @xmath370 there exists an integer @xmath39 for which @xmath371 , hence @xmath372 has finite order by lemma [ prop : z21 ] .
the converse is proposition [ prop : z3 ] ( ii ) . @xmath137
let @xmath2 be a twistor space on @xmath0 , and @xmath345 be the natural square root of the anti - canonical bundle over @xmath2 , which is known to exist on any twistor space @xcite .
following poon @xcite , we call @xmath345 the _ fundamental line bundle _ , and the associated linear system @xmath373 the _
fundamental system_. basic properties of the fundamental line bundle are @xmath374 where @xmath315 is a fibre of the twistor projection @xmath375 which is called a _
real twistor line _ , and @xmath376 is a natural anti - holomorphic involution called the _ real structure_. by the works of poon @xcite , kreussler and kurke @xcite , when @xmath377 , for any twistor space @xmath2 over @xmath0 whose self - dual metric is of positive scalar curvature , we have @xmath378 , and the structure of these twistor spaces is well understood through the rational map associated to the fundamental system @xmath373 . in particular , all of these twistor spaces are moishezon .
let now @xmath2 be a twistor space on @xmath0 , @xmath10 , and assume that @xmath379 .
in this situation @xmath373 needs to contain an irreducible divisor , because otherwise there would exist two pencils , @xmath380 and @xmath381 , of divisors of degree one that define a surjective rational map to @xmath382 and for which @xmath383 is a fundamental divisor .
but then the image of @xmath373 would be at least two - dimensional and bertini s theorem implies that not all elements in @xmath373 could be reducible .
this allows us to apply the results of @xcite .
first , by ( * ? ?
* theorem 3.6 ) , if @xmath378 , we always have the equality @xmath384 , and @xmath2 has to be a so - called lebrun twistor space @xcite , the structure of which is also well - understood .
in particular , such twistor spaces are moishezon .
second , by ( * ? ? ?
* theorem 3.7 ) , if @xmath385 and @xmath386 , then @xmath2 has to be one of the twistor spaces investigated in @xcite , and they are again moishezon .
third , if @xmath385 and @xmath11 , then @xmath2 is either one of the moishezon twistor spaces studied in @xcite , or a twistor space that satisfies @xmath17 .
the former happens exactly when @xmath387 ( ( * ? ? ?
* proposition 2.4 ) ) , and if the latter is the case , the morphism @xmath388 associated to the net @xmath373 is an elliptic fibration which is an algebraic reduction of @xmath2 .
thus the basic structure of @xmath2 is also well - understood if @xmath385 . for the rest of this paper
we let @xmath2 be a twistor space over @xmath0 @xmath389 and suppose @xmath390
. then general members of the fundamental system @xmath373 are irreducible , since otherwise we readily have @xmath378 .
this implies that the self - dual metric on @xmath0 has positive scalar curvature , see ( * ? ? ?
* proposition 2.4 ) .
let @xmath391 be a smooth fundamental divisor and recall that @xmath392 because @xmath2 is simply connected .
as we assume @xmath393 , the standard exact sequence @xmath394 implies @xmath395 .
this means that the anti - canonical system @xmath396 consists of a single member , say @xmath53 .
in particular , we have @xmath89 . from the surjectivity of the restriction map
@xmath397 we have @xmath398 by a theorem of pedersen and poon @xcite , any real irreducible member @xmath391 is non - singular and obtained from @xmath399 by blowing up @xmath400 points .
on such a surface , the anti - canonical curve @xmath53 is real ( i.e. @xmath401 ) since @xmath52 and so @xmath402 are real .
moreover we have the following result on the structure of the base curve @xmath53 .
[ prop : cycle ] let @xmath375 , @xmath10 , be a twistor space satisfying @xmath403 , and let the curve @xmath53 be the base locus of the pencil @xmath373 , as above .
if @xmath53 is non - singular , it is an elliptic curve .
if @xmath53 is singular , it is a cycle of rational curves which is of the form @xmath404 for some @xmath405 , where @xmath406 means @xmath407 , and in the presentation two components intersect iff they are adjacent , or they are @xmath408 and @xmath409 .
recall that @xmath410 for each smooth real @xmath391 .
if @xmath53 is non - singular , the adjunction formula immediately implies that @xmath53 is an elliptic curve .
if @xmath53 is singular it is a cycle of rational curves , by ( * ? ? ?
* proposition 3.6 ) .
if @xmath53 arises from item ( i ) or ( iii ) in ( * ? ? ?
* proposition 3.6 ) , it consists of conjugate pairs of rational curves , as required in . if @xmath53 arose from item ( ii ) in ( *
* proposition 3.6 ) it would have exactly two irreducible components , one of them a real twistor line .
because a real twistor line in @xmath52 generates a pencil , we would readily have @xmath411 , in contradiction to our observations just after the exact sequence .
therefore @xmath53 is a cycle of rational curves as described in .
@xmath137 in the sequel , for simplicity of notation , we often write @xmath412 for @xmath413 and @xmath414 for @xmath408 for components of the cycle
. this cycle will be significant throughout our proof of the main result .
another important property of twistor spaces satisfying @xmath415 , which we next explain , concerns reducible members of the pencil @xmath373 .
the cycle can be split into connected halves in exactly @xmath103 ways . for example , if @xmath416 , the possibilities are : @xmath417 the following proposition , which was proved in @xcite , implies that if the base curve is singular , these subdivisions are nicely realised by reducible members of the pencil @xmath373 .
[ prop : d1 ] ( ( * ? ? ?
* proposition 3.7 ) ) if the base curve of the pencil @xmath373 is a cycle of rational curves as in , then @xmath373 has exactly @xmath103 reducible members .
moreover each of them is real and of the form @xmath418 ( @xmath128 ) , where @xmath419 and @xmath420 are non - singular irreducible divisors satisfying @xmath421 .
furthermore , the divisor @xmath418 splits the cycle @xmath53 into halves in the following manner : * if @xmath422 denotes the real twistor line joining the two points @xmath423 and @xmath424 , then @xmath425 , * the intersections @xmath426 and @xmath427 are connected .
note that all the reducible fundamental divisors @xmath418 are singular along the twistor line @xmath428 .
therefore , all smooth fundamental divisors are automatically irreducible .
we will also need the following property of the zariski decomposition of the cycle @xmath53 .
[ prop : indep ] in the situation of proposition [ prop : cycle ] , let @xmath52 be a smooth member of the pencil @xmath373 . then , the degree of @xmath52 ( see ) and the zariski decomposition of the cycle @xmath410 are independent of the choice of @xmath52 .
let @xmath391 be any smooth member , which is irreducible as we have seen above , and let @xmath410 the base locus of the pencil @xmath373 .
for the self - intersection numbers in @xmath52 of components of the cycle @xmath53 we have @xmath429 hence these self - intersection numbers in @xmath52 are independent of the choice of the smooth member @xmath52 .
obviously the intersection numbers between different components of @xmath53 are independent of the choice of @xmath52 as well .
let @xmath66 be the zariski decomposition of the cycle @xmath53 regarded as a curve in @xmath52 .
then @xmath55 is nef as a divisor in @xmath52 . in particular
@xmath430 for any index @xmath65 . as the intersection number @xmath431
is determined by the coefficients of @xmath55 and the intersection numbers @xmath432 , it is independent of the choice of @xmath52 and it follows that @xmath55 is nef also in any other smooth member of @xmath373 .
similarly , @xmath56 is negative definite not only in @xmath52 but also in any other smooth member of @xmath373 . by the same reason
, we have @xmath135 for each @xmath102 which is included in @xmath56 , not only in @xmath52 but also in any other smooth member of @xmath373 . thus all the properties ( i ) , ( ii ) and
( ii ) that characterise the zariski decomposition are satisfied for @xmath433 in all smooth members of the pencil @xmath373 . therefore
, the zariski decomposition of the cycle @xmath53 is independent of the choice of a smooth member @xmath391 .
this implies that the self - intersection number @xmath434 is also independent of the choice of @xmath52 .
hence we obtain the proposition .
@xmath137 note that , even though the zariski decomposition @xmath66 is independent of @xmath52 , it is possible that the divisor @xmath372 in @xmath435 does depend on @xmath52 .
the reason is that the restriction @xmath372 actually is @xmath436 and like @xmath437 , the normal bundle of @xmath53 in @xmath52 , this may depend on @xmath52 .
the independence of the zariski decomposition just means that the rational coefficients of @xmath55 and @xmath56 do not depend on @xmath52 .
since the works of poon @xcite and campana @xcite , it has long been well realised that the algebraic dimension @xmath3 of a twistor space @xmath2 on @xmath0 is closely related to the anti - kodaira dimension of a smooth member of the system @xmath373 .
actually , if @xmath52 is a smooth member of @xmath373 , the standard exact sequence @xmath438 and the equality @xmath439 imply the inequality @xmath440 does equality hold in when @xmath2 satisfies @xmath403 ?
for instance , if at least one smooth member @xmath52 of the pencil @xmath373 satisfies @xmath125 , we clearly have the equality @xmath441 .
the following proposition shows that the same is true if some @xmath442 satisfies @xmath97 .
[ prop : ak2 ] if at least one smooth member @xmath52 of the pencil @xmath373 satisfies @xmath97 , then equality holds in , and the twistor space @xmath2 is moishezon .
let @xmath442 be smooth with @xmath97 .
if the base curve @xmath53 of the pencil @xmath373 is non - singular , we readily obtain @xmath443 from @xmath444 .
hence @xmath53 is singular , and by proposition [ prop : cycle ] , @xmath53 is a cycle of rational curves on @xmath52 as in .
let @xmath66 be the zariski decomposition of @xmath53 on @xmath52 .
for the degree of @xmath52 we then have @xmath98 by proposition [ prop : degree ] .
since the degree is independent of the choice of a smooth member @xmath52 of @xmath373 by proposition [ prop : indep ] , we obtain that @xmath445 for any smooth member @xmath446 . by proposition [ prop :
degree ] , this implies @xmath447 for any smooth @xmath446 .
once this is obtained , the equality @xmath448 follows from a general estimate of the algebraic dimension from below in the case of a fibre space .
we refer to ( * ? ? ?
* proposition 4.1 ) and ( * ? ? ?
* corollary 4.3 ) for details .
@xmath137 in order to determine the algebraic dimension of @xmath2 if @xmath403 , we are left with the case where the pencil @xmath373 contains a smooth member @xmath52 satisfying @xmath370 .
this is well understood for the case @xmath11 , which we next explain in order to clarify the difference to the case @xmath1 .
let @xmath2 be a twistor space on @xmath13 satisfying @xmath449 and let @xmath53 be the base locus of @xmath373 .
suppose that @xmath2 has a smooth fundamental divisor of anti - kodaira dimension one .
using , this implies @xmath450 .
as we assume @xmath11 , we can apply ( * ? ? ?
* theorem 6.2 ) to conclude that @xmath345 is nef .
therefore , @xmath451 is nef for each smooth @xmath442 , i.e. @xmath312 in the zariski decomposition of @xmath53 on @xmath52 . as @xmath452 , corollary [ cor : finord ] shows now that @xmath370 if and only if the line bundle @xmath453 is of finite order in @xmath138 . because @xmath454 this order does not depend on @xmath52 .
hence the anti - kodaira dimension of smooth members of @xmath373 is constant . from this , exactly like at the end of the proof of proposition [ prop : ak2 ] , we obtain the following result . [ prop : a2 ] let @xmath2 be a twistor space over @xmath13 satisfying @xmath403 .
suppose that there exists a smooth member @xmath52 of the pencil @xmath373 satisfying @xmath370 . then @xmath17 .
we should mention that a much more concrete result was obtained in ( * ? ? ?
* theorem 3.4 ) without using the general estimate on the algebraic dimension of a fibred space .
namely , if @xmath37 denotes the order of the line bundle @xmath455 in @xmath138 , then @xmath456 , @xmath457 , and the associated morphism @xmath458 provides an algebraic reduction of @xmath2 which is an elliptic fibration . strictly speaking
, the paper @xcite assumes that the base curve @xmath53 is smooth , but the proof equally works even if it is a cycle of rational curves because @xmath455 is trivial on each component of @xmath53 in the situation considered here .
although we are assuming @xmath403 here , the case @xmath459 actually happens , and then @xmath385 , see @xcite .
now we are ready to state the main result of this article .
it says that , in contrast to the case @xmath11 , when @xmath1 and @xmath370 for some @xmath442 , equality in never holds .
[ thm : main ] let @xmath1 and @xmath2 be a twistor space on @xmath0 satisfying @xmath403 .
suppose there exists a smooth member @xmath52 of the pencil @xmath373 such that @xmath370 , then @xmath441 .
this means that when @xmath1 , if a twistor space @xmath2 on @xmath0 satisfies @xmath403 , a general member @xmath52 of the pencil @xmath373 will satisfy either @xmath460 or @xmath461 .
the algebraic dimension of such a twistor space can not be equal to @xmath198 and smooth fundamental divisors with @xmath370 can only occur if @xmath441 .
the proof of theorem [ thm : main ] will be completed at the end of this section . in preparation for this proof
, we first note that under the assumptions of theorem [ thm : main ] the base curve @xmath53 of @xmath373 can not be smooth and moreover @xmath105 holds in .
[ prop : cycle2 ] let @xmath2 and @xmath52 be as in theorem [ thm : main ] , and @xmath53 be the unique anti - canonical curve on @xmath52 . then @xmath53 is a cycle of rational curves as in with @xmath105 . by proposition [ prop : cycle ] , the base curve @xmath53 is either a smooth elliptic curve or a cycle of rational curves as in .
if @xmath53 is smooth , from @xmath462 we easily obtain @xmath463 for any @xmath67 , but then @xmath464 . therefore @xmath53 can not be smooth . if @xmath53 is as in with @xmath465 , then @xmath466 , and @xmath467 as @xmath53 is a cycle . together with @xmath468 ,
this means @xmath469 .
hence the intersection matrix for the cycle @xmath53 becomes @xmath470 if @xmath1 , this is negative definite .
hence @xmath124 in the zariski decomposition @xmath66 of @xmath53 , which implies @xmath125 , see lemma [ lem : pzero ] ( i )
. therefore @xmath105 .
@xmath137 let @xmath471 and @xmath53 be as in theorem [ thm : main ] and proposition [ prop : cycle2 ] and let @xmath472 be the blow - up with centre @xmath53 .
the space @xmath473 has singularities , but we discuss this later .
let @xmath61 and @xmath474 ( @xmath128 ) be the exceptional divisors over the curves @xmath102 and @xmath406 , respectively .
write @xmath475 for the sum of all exceptional divisors .
any two distinct smooth members @xmath476 and @xmath477 of the pencil @xmath373 intersect transversally along the cycle @xmath53 in the sense that @xmath478 as a divisor on @xmath476 . therefore ,
the blow - up @xmath472 composed with the rational map @xmath479 associated to the pencil @xmath373 is a morphism @xmath480 the fibres of which are the strict transforms of the members of the pencil @xmath373 .
this is nothing but an elimination of the indeterminacy locus of the rational map @xmath479 .
we will use the same letters to denote divisors in @xmath2 and their strict transforms in @xmath473 .
recall from proposition [ prop : d1 ] that the pencil @xmath373 has exactly @xmath103 reducible members @xmath418 , and that @xmath105 by proposition [ prop : cycle2 ] .
let @xmath481 , @xmath128 , be the point over which the reducible fibre @xmath482 is lying .
evidently we have the basic relation @xmath483 for any index @xmath65 , the exceptional divisor @xmath61 comes with two fibrations @xmath484 and @xmath485 .
both are clearly @xmath486-bundles , and fibres of different fibrations intersect transversally at one point .
hence we obtain an isomorphism @xmath487 for any @xmath65 .
the same is true for the real conjugate divisor @xmath474 from reality . since the centre @xmath53 of the blow - up @xmath488 is singular , being a cycle of rational curves , the space @xmath473 has ordinary double points over the singularities of the cycle @xmath53 .
concretely , these are exactly the points where the four divisors @xmath489 meet and the real conjugate points of these , see figure [ fig1 ] where these points for @xmath490 are indicated by black circles .
thus @xmath473 has precisely @xmath491 ordinary double points in total .
since each ordinary double point admits two small resolutions , the number of simultaneous small resolutions of all the double points , which are compatible with the real structure , is @xmath492 in total .
let @xmath493 be any one of these small resolutions .
we denote by @xmath494 the composition @xmath495 of the small resolution and the blow - up .
we continue to use the same letters @xmath61 and @xmath474 to denote the strict transforms of the exceptional divisors in @xmath32 .
also , @xmath28 denotes the total sum @xmath496 in @xmath32 .
we write @xmath497 for the composition @xmath498 . from
we obtain the relation @xmath499 and @xmath26 has reducible fibre @xmath482 over the point @xmath500 , @xmath128 .
if we regard the cycle @xmath53 as a curve on a real smooth member of the pencil @xmath373 , it has a zariski decomposition with @xmath55 and @xmath56 both real @xmath501 by proposition [ prop : indep ] , this zariski decomposition is independent of the choice of the smooth member . as in section
[ s : z ] , let @xmath36 be the smallest integer for which @xmath122 is integral , and write @xmath502 for later use , we use these coefficients @xmath267 and define a @xmath57-divisor @xmath503 on @xmath32 by the equation @xmath504 similarly we define a @xmath57-divisor @xmath505 on @xmath32 by the equation @xmath506 thus @xmath503 and @xmath505 may be regarded as enlargements of the nef part and the negative part , respectively , of the zariski decomposition to divisors in @xmath32 .
the key feature of these divisors is that if we restrict to a smooth fibre of the projection @xmath507 , we obtain the zariski decomposition .
because we assume @xmath301 for a given @xmath442 , we obtain from proposition [ prop : degree ] and lemma [ lem : pzero ] ( i ) that @xmath126 and @xmath181 in @xmath52 . because of proposition [ prop : indep ] the same is true for all smooth fundamental divisors .
in particular , we can apply proposition [ prop : z2 ] to get @xmath182 for all @xmath65 and @xmath236 for at least one @xmath65 .
moreover , by proposition [ prop : z2 ] ( iv ) , as @xmath1 , the @xmath103 integers @xmath508 can not all be equal to each other . continuing to write @xmath509
, it follows that there is an index @xmath65 for which @xmath510 holds . after a cyclic permutation of the indices we may assume that @xmath511 . for arbitrary integers
@xmath512 we define a line bundle @xmath513 on @xmath32 by @xmath514 [ prop : rest1 ] let @xmath476 be any smooth fibre of the fibration @xmath497 , and identify the pair @xmath515 in @xmath516 with @xmath517 in @xmath2 by the map @xmath494 . then for any @xmath518 and @xmath519 , we have @xmath520 recall that @xmath521 denotes the group of those line bundles on @xmath53 that are of degree zero on each component of @xmath53 . by the identifications @xmath522 and @xmath523 induced by the birational morphism @xmath25 , we obtain that the restriction of @xmath513 to @xmath476 is isomorphic to @xmath524 here we have disposed of @xmath525 since @xmath476 is a fibre of @xmath26 . because @xmath126 on @xmath476 , lemma [ lem : pzero ] ( ii ) implies that @xmath526 , hence so is @xmath527 .
@xmath137 we now take a closer look at the ordinary double point @xmath528 in @xmath473 . without loss of generality , we may suppose that @xmath529 and @xmath530 are distinguished by the property that the intersection @xmath531 is a point , see figure [ fig1 ] .
depicted for @xmath416 .
the bold lines in the right figure mark the cycle @xmath532 that is defined below . ]
the small resolution @xmath493 blows up one of the pairs @xmath533 or @xmath534 . because there is no essential difference between these two
, we may suppose that the pair @xmath533 is blown up .
we denote the exceptional curve over the point @xmath531 by @xmath535 .
since @xmath488 blows up the cycle @xmath53 , we can not speak about strict transforms of the components @xmath102 of @xmath53 into @xmath32 .
but , if we specify a fibre of @xmath497 , we can .
in particular , we denote the strict transforms of @xmath139 and @xmath536 in the fibre @xmath537 by @xmath538 these are identified with @xmath102 and @xmath406 , respectively , by the birational morphism @xmath494 .
the first index , @xmath179 , indicates that the curves are included in the fibre @xmath539 , and the second index reflects that they correspond to @xmath102 and @xmath406 respectively . unlike in the original space @xmath2 ,
the union of all the curves is not a cycle because the exceptional curves @xmath535 and @xmath540 are inserted between @xmath541 and @xmath542 , and between @xmath543 and @xmath544 , respectively , see figure [ fig1 ] .
hence the union of all the curves consists of two connected chains @xmath545 and by adding @xmath535 and @xmath540 we get a cycle of rational curves , which we denote by @xmath532 .
we clearly have @xmath546 in the sense that @xmath532 is the restriction of the divisor @xmath28 to the fibre @xmath537 or , equivalently , @xmath532 is an element of the linear system @xmath547 on the surface @xmath28 .
the intersection numbers of the line bundle @xmath513 with the components of the cycle @xmath532 will play an essential role in the subsequent proofs . [
lemma : int1 ] we have @xmath548 and @xmath549 let @xmath476 be any smooth fibre of @xmath497 . if @xmath550 , the curve @xmath551 is homologous in @xmath61 to @xmath552 .
therefore , using proposition [ prop : rest1 ] , we have @xmath553 the case @xmath554 requires more attention .
dropping components that are disjoint from @xmath542 , see figure [ fig1 ] , we first obtain @xmath555 further , as @xmath556 is one point and the intersection is transverse , we have @xmath557 .
next , @xmath558 since @xmath559 and the intersection is transverse .
moreover , since the pairs @xmath560 in @xmath32 and @xmath561 in the original twistor space @xmath2 are isomorphic , the self - intersection number @xmath562 in @xmath563 is equal to @xmath564 calculated in @xmath565 .
we let @xmath566 , calculated in @xmath567 . by the adjunction formula and
because @xmath568 on @xmath2 , we have @xmath569 by lemma [ lem : pzero ] ( ii ) we have @xmath570 on the original surface @xmath567 of which we assumed @xmath301 .
thus , we have the relation @xmath571 with which we obtain @xmath572 therefore , from , we get @xmath573 for the remaining intersection number we notice that the curve @xmath574 , regarded as a curve in the surface @xmath575 , is homologous to the fibre @xmath576 of the projection @xmath577 . because @xmath578 and @xmath579 is zero , we obtain @xmath580 and follows from .
@xmath137 let @xmath581 be an integer and recall that we have chosen notation so that @xmath582 . from for @xmath554
we then obtain @xmath583 .
using for @xmath584 and reality it follows now that the two chains @xmath585 are contained in the base locus of the linear system @xmath586 .
this will be a stepping stone for the following stronger statement , which will play a key role in the proof of the main theorem .
[ prop : keyvan ] for any @xmath587 and @xmath581 , we have @xmath588 in particular , if @xmath581 the divisor @xmath28 is a fixed component of the linear system @xmath586 on @xmath516 .
recall that the cycle of rational curves @xmath532 belongs to the linear system @xmath589 on @xmath28 and that @xmath590 . therefore , multiplication by a section of @xmath591 with zero locus @xmath532 provides an exact sequence @xmath592 to first show that @xmath593 in the induced exact sequence @xmath594 we let @xmath595 be a non - zero section and note that @xmath596 .
suppose that @xmath330 identically vanishes on a component of @xmath28 .
then since , for each @xmath65 the degree of the line bundle @xmath513 on a general fibre of @xmath597 is zero by proposition [ prop : rest1 ] , we obtain that @xmath330 vanishes identically on @xmath28 .
hence @xmath330 can not vanish identically on any component of @xmath28 .
therefore we can consider the effective cartier divisor @xmath598 on @xmath28 , and the restriction @xmath599 is a curve on @xmath61 for any @xmath65 . from the intersection numbers we have seen above that the divisor @xmath598 includes the two connected chains for @xmath581 and all @xmath587 .
hence the curve @xmath600 passes through the point @xmath601 , see figure [ fig1 ] .
this point is not on @xmath542 , because @xmath541 and @xmath542 are disjoint curves in @xmath602 .
if the divisor @xmath600 had a component other than @xmath535 , containing this point , then we would have @xmath603 by proposition [ prop : rest1 ] , since the fibre of @xmath604 through this point is @xmath605 and so @xmath600 would have to intersect any fibre of the projection @xmath606 .
therefore such a component does not exist .
hence the divisor @xmath600 , and so @xmath598 , includes @xmath535 .
the same argument shows that @xmath607 .
we have seen before that @xmath598 includes all the other components @xmath551 and @xmath608 of @xmath609 and so we obtain @xmath610 , which implies @xmath611 .
this shows that @xmath612 . from the exact sequence we obtain now for all @xmath581 and all @xmath587 that
@xmath613 on the other hand , using and the projection formula for @xmath614 , we obtain @xmath615 for any coherent sheaf @xmath616 on @xmath617 and for sufficiently large @xmath191 , @xmath618 .
taking @xmath191 so that this holds for @xmath619 , we obtain from that @xmath620 for all @xmath587 and all @xmath581 .
@xmath137 by propositions [ prop : rest1 ] the restriction of @xmath621 to @xmath622 gives an element @xmath623 for each value of @xmath624 for which the fibre @xmath625 is smooth .
[ prop : lnu2 ] assume that @xmath626 has the same finite order @xmath37 for all smooth fibres @xmath625 .
then , for any @xmath587 and @xmath39 , we have @xmath627 recall that @xmath628 by proposition [ prop : keyvan ] the divisor @xmath28 is a fixed component of @xmath629 and so @xmath630 let @xmath476 be a general fibre of @xmath497 , and identify the pair @xmath515 with the pair @xmath517 included in the original space @xmath2 by the birational morphism @xmath494 . because we assumed @xmath626 to have finite order , we can use proposition [ prop : z3 ] ( iv ) to obtain @xmath631 on @xmath476 . as this is valid for any smooth fibre of @xmath26 , we obtain that the divisor @xmath632 on @xmath32 is a fixed component of the system @xmath633 . hence , from and using definition ,
we obtain @xmath634 using induction on @xmath39 we finally obtain @xmath635 as required .
we are now ready to prove our main result .
_ proof of theorem [ thm : main ] .
_ if the order of the restriction @xmath626 of @xmath621 to @xmath636 is not the same for all smooth fibres @xmath625 , the algebraic dimension of @xmath2 is equal to @xmath179 . to see this , note that the set of points of finite order in @xmath521 is discrete and that we have a continuous function @xmath637 which is defined on the open set in @xmath617 over which @xmath638 has smooth fibres .
if the order of @xmath626 is not constant , the image of the above map contains an element of infinite order . by proposition [ prop : degree ] and corollary [ cor : finord ] this means that there exists a smooth fundamental divisor @xmath639 with @xmath640 . as we have seen from , this implies @xmath441 . it remains to consider the situation in which the order of @xmath626 is the same for all smooth fibres @xmath625 .
let @xmath37 denote this order .
we shall show that this situation is impossible . to derive a contradiction , first note that corollary [ cor : finord ] implies that @xmath301 for each smooth @xmath442 as we have assumed that @xmath626 is of finite order for all smooth fibres @xmath625 . using ( * ? ? ?
* corollary 4.3 ) we obtain that @xmath17 . on the other hand , with the aid of proposition [ prop : lnu2 ] we can show that @xmath441 .
it is interesting to observe that the constancy of the order of @xmath626 will only be used when proposition [ prop : lnu2 ] is applied . here
are the details . for each integer @xmath39 , gives a zariski decomposition in @xmath52 @xmath641 and since the zariski decomposition of the cycle @xmath53 is independent of the smooth member @xmath52 by proposition [ prop : indep ] , this is the zariski decomposition of the divisor @xmath642 on any smooth member of the pencil @xmath373 .
therefore , by proposition [ prop : z1 ] , for any non - singular fundamental divisor @xmath446 , the divisor @xmath643 is a fixed component of the system @xmath644 on @xmath476 .
hence the pull - back @xmath645 to @xmath32 has the divisor @xmath646 as a fixed component ; the divisor @xmath505 was defined in .
we define a line bundle @xmath31 on @xmath32 by @xmath647 where we have used the relation to get the isomorphism .
the line bundle @xmath31 is equipped with a natural real structure coming from that on @xmath345 on @xmath2 .
the morphism @xmath648 provides a bijection between the members of the two linear systems @xmath649 on @xmath2 and @xmath650 on @xmath32 .
moreover , the composition of @xmath651 with the rational map defined by @xmath649 is the rational map which is given by @xmath650 .
because @xmath652 by proposition [ prop : lnu2 ] , we see that the rational map associated to @xmath650 has one - dimensional image , equal to the image of a veronese embedding of @xmath617 .
hence , for all @xmath39 , the rational map defined by @xmath649 has one - dimensional image as well , and this implies that @xmath653 .
this contradicts our previous calculation of the algebraic dimension , hence it is impossible that @xmath626 has finite order not depending on @xmath624 .
in @xcite , the existence of a twistor space on @xmath0 with algebraic dimension two is claimed for any @xmath1 .
the proof of this assertion consists essentially of the following two parts . 1 . for a non - singular surface @xmath52 which admits an anti - canonical cycle , it is shown in ( *
* lemma 3.2 ) that the finiteness of the order of a certain line bundle @xmath654 on a cycle of rational curves implies @xmath370 . with the aid of formula
it can be shown that the pull - back of the line bundle @xmath654 to @xmath52 is isomorphic to the line bundle @xmath655 studied in section 2 .
if a twistor space @xmath2 on @xmath0 , @xmath1 , admits a divisor in @xmath373 which is biholomorphic to the surface @xmath52 in the above part 1 satisfying @xmath370 , then we always have @xmath403 .
it is claimed in the proof of ( * ? ? ?
* lemma 4.3 ) that the isomorphism class of the line bundle @xmath654 from part 1 is the same for generic members of the pencil @xmath373 .
it follows that a generic fundamental divisor @xmath52 in a twistor space @xmath2 as in part 2 above satisfies @xmath370 . using ( * ? ? ?
* proposition 4.1 ) it is then concluded that @xmath17 .
this conclusion is in conflict with our theorem [ thm ] .
each non - singular fundamental divisor @xmath656 ( @xmath657 ) is given as an iteration of blow - ups @xmath658 here , @xmath659 is the blow - up of @xmath660 points , all of which are nodes of an anti - canonical cycle on @xmath399 consisting of @xmath660 components .
the morphism @xmath661 is a composition of blow - ups of which each centre is a node of the anti - canonical cycle .
both @xmath662 and @xmath663 are toric and their isomorphism classes are independent of @xmath657 . on the other hand , @xmath664 blows up @xmath660 points , where each centre belongs to the strict transforms of the @xmath660 exceptional curves of @xmath659 which are disjoint to each other .
thus the variation of the complex structure on @xmath656 is entirely encoded in the variation of the @xmath660 points blown up under @xmath665 .
it is important to keep in mind that the sequence is constructed for each non - singular @xmath656 individually , and there is no natural way to identify the surface @xmath662 ( and @xmath663 and @xmath399 also ) for different choices of @xmath657 .
let @xmath53 be the anti - canonical cycle on @xmath656 , which is the base curve of the pencil @xmath373 , and @xmath666 .
the argument that the order of the line bundle @xmath667 does not depend on the surface @xmath52 in the fundamental pencil is built in an essential way on the assertion that the pull - back @xmath668 does not depend on @xmath669 .
this independence is based on the claim that the morphism @xmath670 does not depend on @xmath669 .
we now derive a contradiction from the assumption that @xmath670 is the same for two nearby values of @xmath669 for which @xmath671 is a real smooth member of the fundamental pencil .
this confirms that there is a gap in the proof of ( * ? ? ?
* lemma 4.3 ) on line 10 , page 1105 .
let @xmath52 and @xmath476 be two real irreducible fundamental divisors on a twistor space @xmath2 over @xmath672 such that both are obtained be a sequence of blow - ups from @xmath382 as in .
the linear system of twistor lines on each of these smooth surfaces gives , see @xcite , a morphism with target @xmath617 and we can choose the morphisms in so that their composition with the first projection @xmath673 is the morphism given by the pencil of twistor lines .
we now identify the surfaces @xmath662 , @xmath663 and @xmath382 obtained by blowing down some curves on @xmath52 with those obtained by blowing down some curves on @xmath476 .
as mentioned above , there is no natural way to do so . from proposition [ prop : d1 ]
we can deduce that the two component of the cycle @xmath53 that are met by real twistor lines in a real fundamental divisor @xmath671 do not change as long as @xmath669 varies within one connected component of the set of real irreducible fundamental divisors , which is the complement of a finite set of points on a circle .
we assume that @xmath52 and @xmath476 belong to such a connected component and denote the morphisms @xmath665 for these two surfaces by @xmath674 and @xmath675 , respectively .
we let @xmath153 be the composition @xmath676 and define @xmath677 similarly .
both restrictions @xmath678 and @xmath679 have the same image @xmath680 . because @xmath681 is the same for both surfaces @xmath52 and @xmath476 , the assumption that @xmath682 implies that the two compositions @xmath683 and @xmath684 coincide .
let now @xmath685 be a real twistor fibre , then @xmath686 is a set of two conjugate points . consider the fibre @xmath687 of @xmath688 over the point @xmath689 .
as @xmath688 has only finitely many reducible fibres and the real twistor lines in @xmath52 and @xmath476 meet the same components of @xmath53 , we can choose @xmath315 so that @xmath690 is irreducible . because , by assumption , @xmath691 , we have @xmath692 as well .
however , as there do not exist two real twistor fibres in @xmath2 through the point @xmath693 and @xmath694 does not contain a real twistor fibre , @xmath690 can not be a real member of the pencil @xmath51 in @xmath476 .
hence , @xmath690 and its conjugate are two different members of the same pencil on @xmath476 and they intersects at least at the two points @xmath693 and @xmath695 .
but this is absurd as the irreducible curve @xmath690 has self - intersection number @xmath129 on @xmath476 .
this contradiction shows that @xmath670 can not be the same for any two values of @xmath669 for which @xmath671 is real .
a detailed analysis of the argument in @xcite reveals that the core problem is whether for @xmath696 the automorphism @xmath697 extends to an automorphism @xmath698 , or more precisely , to an isomorphism between neighbourhoods of @xmath699 .
this seems to have been overlooked in @xcite .
a priori there is no reason for this to hold , and our theorem [ thm ] shows that for generic @xmath700 , such an extension can not exist and even if some member in the pencil @xmath373 satisfies @xmath87 , the general member of the pencil needs to satisfy @xmath701 .
f. campana , _ the class @xmath702 is not stable by small deformations _ , math .
* 229 * ( 1991 ) 1930 .
f. campana , b. kreuler , _ a conic bundle description of moishezon twistor spaces without effective divisor of degree one _ , math . z.
* 229 * ( 1998 ) 137162 .
f. campana , b. kreuler , _ existence of twistor spaces of algebraic dimension two over the connected sum of four complex projective planes _ , proc .
* 127 * ( 1999 ) 26332642 .
n. hitchin , _ khlerian twistor spaces _ , proc .
london math .
( 3 ) * 43 * ( 1981 ) 133 - 150 .
n. honda and m. itoh , _ a kummer type construction of self - dual metrics on the connected sum of four complex projective planes _ , j. math .
japan * 52 * ( 2000 ) 139 - 160 .
b. kreuler , _ on the algebraic dimension of twistor spaces over the connected sum of four complex projective planes _ , geom .
dedicata * 71 * ( 1998 ) 263 - 285 .
b. kreuler , _ twistor spaces with a pencil of fundamental divisors , _ doc.math.j . * 4 * ( 1999 ) 127 - 166 .
c. lebrun , _ explicit self - dual metrics on @xmath703 _ , j. differential geom .
* 34 * ( 1991 ) 223253 .
h. pedersen , y. s. poon , _ self - duality and differentiable structures on the connected sum of complex projective planes _
* 121 * ( 1994 ) 859 - 864 .
y. s. poon , _ compact self - dual manifolds of positive scalar curvature _
, j. differential geom . *
24 * ( 1986 ) 97132 .
y. s. poon , _ algebraic dimension of twistor spaces _
ann . * 282 * ( 1988 ) 621 - 627 . | we show that the algebraic dimension of a twistor space over @xmath0 can not be two if @xmath1 and the fundamental system ( i.e.the linear system associated to the half - anti - canonical bundle , which is available on any twistor space ) is a pencil .
this means that if the algebraic dimension of a twistor space on @xmath0 , @xmath1 , is two , then the fundamental system either is empty or consists of a single member .
the existence problem for a twistor space on @xmath0 with algebraic dimension two is open for @xmath1 . |
the sunspot number is the solar index most frequently used in the study of long - term solar activity variations and , even more so , in solar - terrestrial relation studies .
the relative sunspot number was defined by rudolf wolf @xcite as @xmath0 where @xmath1 is the number of sunspot groups , @xmath2 is the total number of ` spots ' in all the groups on the visible disk , and @xmath3 is a scale factor to bring the number on to wolf s scale ( thus @xmath4 for wolf himself ) .
it would seem that @xmath1 and @xmath2 should be uniquely determined simply by counting and that a @xmath3-factor would not be necessary .
however , different observers - even using the same instrument - may differ in how they overcome variable seeing and arrive at different numbers of groups .
even more so for the number of spots , where the very definition of what should be counted as a spot may vary from observer to observer .
issues here are the distinction between spots with penumbrae and pores without , the treatment of spots that touch each other , different weighting according to size , whether to count all spots or only the larger ones , observer acuity and snellen ratio , etc .
different observers have different answers to , preferences of , and opinions about these issues . with an appropriate @xmath3-factor
their observed counts are , presumably , reduceable to the wolf - scale .
there is some confusion as to the precise meaning of the reduction factor . strictly speaking
, it should only apply to the standard instrument : 8 cm refractor at magnification 64 . to compensate for a larger or smaller telescope an additional factor should be employed . in practice
this is too cumbersome and the additional factor is folded into the @xmath3-factor . for a detailed discussion of these issues
see @xcite and @xcite .
the f10.7 cm ( 2.8 ghz ) solar index , introduced by covington in 1947 is generally viewed as an excellent index of solar activity @xcite .
the basic minimum level of emission around 67 sfu ( solar flux unit = @xmath5 w m@xmath6 hz@xmath7 ) is presumed to come from the quiet background sun .
an @xmath8-component due to solar activity on time scales longer than those of flares is fashioned into the f10.7 index measured and maintained by the solar radio monitoring programme operated jointly by the national research council and natural resources canada ( http://www.spaceweather.gc.ca/sx-eng.php ) .
routine observations at 1.0 , 2.0 , 3.75 , and 9.4 ghz straddling the 2.8 ghz frequency of the canadian series have been made in japan since the 1950s ( ftp://solar.nro.nao.ac.jp/pub/norp/data ) .
the two microwave datasets [ suitably scaled ] compare favorably with one another and testify to the stability ( to within the accuracy of the measurements ) of the calibration of both @xcite .
in a short 1971 paper , max waldmeier @xcite pointed out that the zrich standard scale [ of the relative sunspot numbers ] has never been calibrated in an objective way " .
he went on to note that the close correlation between monthly , and especially yearly , means of the solar microwave emission at 10.7 cm wavelength and the sunspot numbers yields a possibility of an objective calibration of the scale of the relative sunspot numbers .
figure [ f - relation ] shows the tight relationship [ linear for sunspot number greater than 25 ] deduced by waldmeier for the interval 1947 - 1970 ( black dots ) .
he remarks that as long as this relation holds , the zrich series of sunspot - numbers may be considered to be homogeneous .
if this relation should be subject to changes in the time to come , then the reduction factor used hitherto ought to be changed in such a way that the old r - f relation is reestablished " .
figure [ f - relation ] also shows the relation since 1996 derived from the international sunspot number as determined by sidc ( red dots , http://sidc.oma.be/data/yearssn.dat ) .
the data for the intervening interval 1971 - 1995 are shown as gray dots and open red squares .
it is clear that the recent sunspot numbers no longer follow the relationship found by waldmeier and that therefore , perhaps ,
the reduction factor used hitherto ought to be changed in such a way that the old r - f relation is reestablished " . on the other hand
it is also possible that the sunspot number as currently defined simply is no longer a suitable measure of solar activity , given the progressive discrepancy with the f10.7 flux .
a similar conclusion was reached by @xcite and @xcite based on monthly values .
one can speculate that the reason for this is the recent reduced visibility of sunspots due to diminished contrast to the surrounding photosphere reported by @xcite on account of weaker magnetic field and increased temperature .
recent measurements appear to confirm those trends as seen in figure [ f - umbral ] [ livingston , personal communication ] .
should such deviations from ` normal ' observed sunspot activity be substantiated in the near future , the question naturally arises whether [ and when ] they might have occurred in the past as well , _
e.g. _ during the maunder minimum , 1645 - 1715 .
i thank bill livingston for his measurements of umbral magnetic field and intensities .
i acknowledge the use of sunspot data from sidc , rwc belgium , world data center for the sunspot index , royal observatory of belgium . and of f10.7 cm flux from national research council and natural resources , canada .
penn , m. j. , and w. livingston ( 2006 ) , temporal changes in sunspot umbral magnetic fields and temperatures , _ astrophys .
journal , _ _
649(1 ) _ , l45-l48 , doi : 10.1068/508345 .
hossfield , c. h. ( 2001 ) , a history of the zurich and american relative sunspot number indices , _
star obs . , _
_ 30 _ , 48 - 53 .
schaefer , b. e. ( 1993 ) , visibility of sunspots , _ astrophys .
journal , _
_ 401 _ , 909 - 919 .
svalgaard , l. , and h. s. hudson ( 2010 ) , the solar microwave flux and the sunspot number , _ soho-23 : understanding a peculiar solar minimum , asp conference series _ ,
eds . : s. r. cranmer , j. t. hoeksema , and j. l. kohl , _ 428 _ , 325 - 328 . tapping ,
k. f. , and d. p. charrois ( 1994 ) , limits to the accuracy of the 10.7 cm flux , _ solar phys . , _ _ 150 _ , 305 - 315 . tapping , k. f. , ( 2010 ) , properties of the sunspot number and the 10.7 cm solar flux activity indices , their interrelationship and unusual behaviour since the year 2000 , _ sorce meeting 2010 , keystone , co. _ http://lasp.colorado.edu/sorce/news/2010sciencemeeting/doc/session6/6.03_tapping_f10.7.pdf .
waldmeier , m. ( 1971 ) , an objective calibration of the scale of sunspot - numbers , _ astron . mitt .
sternwarte zrich , _ _ 304 _ , pp10 , [ also available at http://www.leif.org/research/w-ccciv.pdf ] . wolf , j. r. ( 1856 ) , mittheilungen ber der sonnenflecken , _ astron .
sternwarte zrich , _ _ 2 _ , pp13 .
centerline in the darkest part of 1565 sunspot umbrae ( years 1990 - 2010 ) and of the umbral intensity [ small pink dots ] relative to nearby non - spot photosphere ( livingston , 2010 , personal communication ) .
the larger blue and pink circles show yearly mean values .
, title="fig : " ] | @xcite found a very tight relationship between the f10.7 solar radio flux and the sunspot number and suggested using the flux for an objective calibration of the sunspot number .
he suggested that if this relationship changed later on , the sunspot number should be re - calibrated , assuming that the calibration must have drifted with time .
i repeat his analysis using data up to the present and it is , indeed , clear that the relationship has changed significantly
. this could be due to a drift of the calibration or to a secular change in the visibility of sunspots , or both . |
perhaps the simplest model of noise in a quantum system is that of the isotropic depolarizing channel @xmath2 where with probability probability a quantum state @xmath3 is left untouched while with probability @xmath4 it is mapped to the completely mixed state @xmath5 @xmath6 this channel results in the bloch `` ball '' being compressed isotropically by a factor of .
one can imagine a slightly more complicated noise model whereby the noise compresses the bloch `` ball '' anisotropically along axes which are defined by the basis we choose to work in and it is this setting which we investigate below .
the amounts by which we compress along each axis are called the * compression coefficients * and these form the components of the * compression vector*. we investigate the geometric properties of the set of all compression vectors and in particular we ask `` when does this set form a simplex ? '' these * generalized depolarizing channels * ( also called * anisotropic depolarizing channels * ) form a broad class of channels which can be realized experimentally ( see for example @xcite ) ; in the single qubit case they include the bit - flip and phase - flip channels .
it is worth noting that _ any _ quantum channel ( including the use of a spin chain as a quantum channel @xcite ) can be turned into a generalized depolarizing channel as follows : alice sends one half of a maximally entangled bipartite state down the channel to bob ; the resulting shared state is a mixed bipartite state which is ( excepting very special cases ) not maximally entangled ; alice and bob can now use this shared resource to attempt teleportation @xcite of an unknown @xmath0-dimensional quantum state @xmath3 ; the teleportation protocol then acts as a generalized depolarizing channel on this state @xmath3 @xcite .
this work is organized as follows : section [ sec : problem ] gives some precise definitions and describes in detail the problem we solve .
sections [ sec : p - basis ] , [ sec : gm - basis ] and [ sec : hw - basis ] present the solution to this problem in the pauli , gell - mann and heisenberg - weyl bases respectively ( we prove the conjecture by dixit and sudarshan in section [ sec : p - basis ] ) whilst section [ sec : other - bases ] gives the general solution in an arbitrary basis .
we then discuss changing basis in section [ sec : change - basis ] before showing how to adopt our method of solution to deal with more general channels in section [ sec : more - channels ] .
we conclude in section [ sec : conclusions ] .
throughout this work we let be a basis for matrices which satisfies the following conditions : 1 . @xmath7 2 .
@xmath8 for all @xmath9 3 .
@xmath10 for all @xmath11 abusing terminology slightly ( note that ) , we call such a basis * trace - free * and * trace - orthogonal*. we need not restrict ourselves to trace - orthonormal bases as we can simply divide by @xmath12 when necessary to normalize the basis .
if @xmath3 is the density matrix of any @xmath0-dimensional quantum system , then we may write @xmath13 where we have chosen the normalization such that * if and only if @xmath3 is a pure state * if and only if @xmath3 is a mixed state where the norm of @xmath14 is defined as .
for notational convenience we add an extra @xmath15 component , , which does not affect the value of @xmath16 .
we call the * polarization vector * and @xmath17 the * polarization coefficients * of the state @xmath3 * with respect to the basis @xmath18*. the * bloch `` ball '' * is the set of all polarization vectors corresponding to quantum states .
it is important to note that except for the single qubit case the bloch `` ball '' is not the ball of unit radius , but rather a convex subset of this ball ( this is because for some vectors lying within the unit ball are not valid polarization vectors as they do not correspond to positive states ) .
we can now define a * ( generalized ) depolarizing channel with respect to the basis @xmath18 * to be a map @xmath19 which satisfies 1 .
@xmath19 is a trace - preserving completely positive map 2 .
@xmath19 compresses the bloch `` ball '' in the following manner : @xmath20 we call the * compression vector * of the channel @xmath19 as it specifies the amount by which @xmath19 compresses the bloch `` ball '' along each axis ; the @xmath21 are called * compression coefficients*. again , for notational convenience , we have added a @xmath15 component ( @xmath22 is the compression coefficient for , so ensures that @xmath19 is trace - preserving ) .
note that for all @xmath23 . to see this ,
let the largest possible magnitude of the polarization coefficient @xmath17 be and let @xmath24 be a state with ( so @xmath24 lies on the boundary of the bloch `` ball '' ) .
if then @xmath25 is not a state ( it lies outside the `` bloch ball '' ) , so as claimed .
it is important to note that the notion of generalized depolarizing channels is highly basis dependent : we _ must _ define such channels with respect to a given basis .
we make the following observation * if then hermiticity of @xmath3 tells us that ( where @xmath26is the complex conjugate of @xmath27 ) . furthermore , since @xmath28 must be a quantum state ( and is therefore hermitian ) , . and note its special cases : * if then and .
( note that @xmath21 can be negative . ) * if @xmath29 is hermitian ( ) then and . *
if @xmath29 is skew - hermitian ( ) then @xmath17 is pure imaginary and .
if we work in a fixed basis it is clear that for each depolarizing channel there is a unique compression vector .
it is natural therefore to ask the question `` which vectors @xmath30 are valid compression vectors corresponding to depolarizing channels @xmath19 ? '' from the definition of a depolarizing channel given above it is clear that the answer to this question is `` a vector @xmath30 is a valid compression vector when the induced map @xmath19 is completely positive and ( @xmath19 is trace - preserving ) '' .
it is clear from above that all compression vectors @xmath30 which induce depolarizing channels @xmath19 lie within the finite region .
however we will see that in general the converse is not true : not all vectors in @xmath31 induce depolarizing channels . to help us
decide which induced maps @xmath19 are completely positive we employ the * choi - jamiolkowski representation * @xmath32 where is the computational basis of our @xmath0-dimensional quantum system .
we will make use of the following theorem and lemma [ thm : cj ] a map @xmath33 is completely positive if and only if @xmath34 is positive .
see @xcite .
[ lem : trace - cj ] if where @xmath35 and @xmath36 are operators ( i.e. matrices ) then .
we can write
. then @xmath37 since the basis @xmath18 is trace - orthogonal , we may write @xmath38 and also @xmath39 since @xmath40 is just a constant , a simple application of lemma [ lem : trace - cj ] allows us to calculate the choi - jamiolkowski representation of the channel @xmath19 @xmath41 using theorem [ thm : cj ] we see that @xmath30 is a compression vector corresponding to a completely positive depolarizing channel @xmath19 when the eigenvalues of @xmath34 are all non - negative . in the following sections we work in different bases and determine which vectors @xmath30 are compression vectors .
in this section we restrict ourselves to a multiple qubit setting ( ) .
we choose the basis @xmath18 to be formed of tensor products of the single qubit pauli matrices : let @xmath23 be the number whose base-@xmath42 representation is @xmath43 then we define the basis matrices @xmath29 by @xmath44 where are the pauli spin matrices .
this basis is trace - free and trace - orthogonal .
note that each @xmath29 is hermitian and so all @xmath17 and @xmath21 are real . in this case
it is a simple matter to find the eigenvectors and eigenvalues of @xmath34 . since @xmath34 is an matrix ( it is a _
superoperator _ ) , we can think of it as being an operator on @xmath45 qubits ( as ) .
let @xmath46 ( where ) be bell states on qubits @xmath47 and .
( note : in more usual notation we have , , and ) .
observe @xmath48 which may be summarized as @xmath49 where @xmath50 it is now a straight forward matter to check that the eigenvectors of the choi - jamiolkowski representation of an induced channel @xmath19 are @xmath51 with corresponding eigenvalues @xmath52 the important thing to notice here is that each @xmath53 is a linear combination of the compression coefficients @xmath21 .
since there are different eigenvectors we have found all the eigenvectors and eigenvalues of @xmath34 . we have therefore shown which vectors @xmath30 induce depolarizing channels @xmath19 namely those for which all eigenvalues of the choi - jamiolkowski representation of @xmath19 are non - negative : for all @xmath54 and @xmath55 .
we are now in a position to prove a conjecture of dixit and sudarshan @xcite , which we present as the following theorem .
[ thm : dix - sud ] when and we work in the pauli basis , the set of all compression vectors forms a simplex in compression space .
since each @xmath53 is linear in the compression coefficients @xmath21 then the equation defines a hyperplane in compression space ( which is a real euclidean vector space of dimension ; it has one dimension for each component of the compression vector excepting the @xmath15 component which we suppress as it is identically equal to @xmath56 ) . since @xmath19 is completely positive if and only if all eigenvalues @xmath53 are non - negative , the hyperplanes must enclose precisely the set of of vectors which induce completely positive depolarizing channels @xmath19 .
in particular the hyperplanes enclose a finite region of compression space and the shape of this enclosed region must therefore be a simplex .
we now prove a small lemma before finding the of the simplex ( which are the depolarizing channels whose compression vectors form the vertices of the simplex ) .
the eigenvalues of the choi - jamiolkowski representation of @xmath19 sum to the system dimension , @xmath0 .
first recall that the sum of the eigenvalues of a matrix is simply the trace of that matrix .
then @xmath57 the extremal channels are @xmath58 first note the identity @xmath59 where @xmath60 we now fix @xmath61 and prove that @xmath62 is an extremal channel . first note that @xmath62 is completely positive ( see for example @xcite ) and apply the above identity to see that @xmath63 it is clear then that @xmath62 is a depolarizing channel whose compression vector has components @xmath64 by combining equations ( [ eqn : lambda - gm ] ) and ( [ eqn : phi - g ] ) we see that @xmath65 it is clear that if , for fixed @xmath61 and for all @xmath23 , @xmath66 then .
we now show that there exist @xmath67 and @xmath68 such that ( [ eqn : fg - condition ] ) holds : * when ( i.e. and for all ) then @xmath69 and so ( [ eqn : fg - condition ] ) holds when @xmath70 * when ( i.e. and for all ) then @xmath71 and so ( [ eqn : fg - condition ] ) holds when @xmath72 * when ( i.e. and for all ) then @xmath73 and so ( [ eqn : fg - condition ] ) holds when @xmath74 and @xmath75 are picked as in ( [ eqn : q_r ] ) and ( [ eqn : p_r ] ) above .
we have now shown that there exist @xmath67 and @xmath68 with .
since @xmath62 is completely positive then all the eigenvalues of @xmath76 are non - negative ; they sum to @xmath0 and we have found one which is equal to @xmath0 ; therefore all other eigenvalues are zero @xmath77 clearly then the compression vector @xmath30 of the map @xmath62 lies on all the hyperplanes except , and so @xmath62 must be an extremal channel .
to finish the proof note that there are @xmath78 vertices of the simplex and there are @xmath78 channels of the form @xmath62 so we have found all the extremal channels .
it is worth pointing out that any compression vector in the simplex can be written as a convex linear combination of the * extremal compression vectors * ( that is , the compression vectors which form the vertices of the simplex ) .
this implies that any depolarizing channel can be written as a convex linear combination of the extremal channels @xmath79 conversely , any channel of this form is a depolarizing channel .
note the following relationship @xmath80 for a single qubit ( ) the compression space has dimension 3 and we see that for a general channel @xmath19 @xmath81 and so we have the following correspondence between extremal channels and compression vectors ( we suppress the @xmath15 component of @xmath82 which is always equal to @xmath56 ) @xmath83 the geometry of these single qubit depolarizing channels is illustrated in figure [ fig : qubit - tetrahedron ]
in the previous section we restricted the dimension of the quantum system to be so that we could employ the pauli basis for multiple qubits . in this section
we choose one possible generalization of the pauli basis , namely the gell - mann basis , which allows us to study systems with arbitrary dimension .
the * gell - mann * basis is defined to be @xmath84 where , and . for notational consistency we identify @xmath85 i.e. . note that this basis is trace - free and trace - orthogonal ( and reduces to the pauli basis when ) . furthermore each @xmath29 is hermitian and so all @xmath17 and @xmath21 are real .
we now attempt to find the eigenvectors and eigenvalues of @xmath86 and begin by defining @xmath87 where and .
then it is a simple matter to check that @xmath88 it is clear therefore that @xmath89 are eigenvectors of @xmath86 and that the corresponding eigenvalues @xmath90 are linear in the compression coefficients @xmath21 .
now , @xmath86 has @xmath78 eigenvalues and we have found of them ; let the remaining @xmath0 eigenvalues be @xmath91 ( ) . in order to establish which vectors in compression space induce depolarizing channels we must now find when the remaining @xmath0 eigenvalues of @xmath92 are non - negative .
since is symmetric for all @xmath23 , @xmath86 is also symmetric and consequently has real eigenvalues . by considering the matrix elements of @xmath86
carefully we see that the only non - zero entries in the two columns indexed by @xmath93 and @xmath94 ( with ) are in the two rows indexed by @xmath95 and @xmath96 .
we may conjugate @xmath86 by a permutation matrix @xmath97 ( to permute the rows and columns ) to form a matrix @xmath98 which has identical eigenvalues to @xmath86 . by repeatedly conjugating by permutation matrices we can block - diagonalize @xmath86 to obtain a matrix @xmath99 which has the following structure @xmath100 & & & \\ & \ddots & & \\ & & [ 2 \times 2 ] & \\ & & & [ k(\phi_{\textbf{\textit{v } } } ) ] \\
\end{array } \right)\ ] ] there are blocks of size ( each of which has two eigenvalues , @xmath101 and @xmath102 , for some @xmath47 and @xmath103 ) and a large block which we call @xmath104 of size ( which has eigenvalues ) . by considering the characteristic equation of @xmath104 we see that @xmath105 where we have defined the * eigenvalue sums * @xmath106 to be @xmath107 the following lemma proves that all the eigenvalues are non - negative precisely when all the eigenvalue sums are non - negative . for all if and only if for all .
one way is trivial : if @xmath108 then @xmath109 .
we prove the converse by contradiction : assume that for all , recall that and note that for any @xmath110 ( this is essentially an inclusion - exclusion argument ) .
but then we may split this sum up into two terms @xmath111 it is clear that these sums over @xmath103 odd and @xmath103 even are both non - negative since each @xmath106 is non - negative and @xmath91 appears to an even power in each term .
but then implies that also , which contradicts the assumption that , and so .
since this argument holds for all we are done . returning to @xmath86 we see from equation ( [ eqn : j(phi ) ] ) that each matrix element consists of a linear combination of the compression coefficients @xmath21 a property which is inherited by @xmath104 .
careful consideration reveals that @xmath106 is a @xmath112-order polynomial in the compression coefficients @xmath113 it is tempting to conclude that is a @xmath112-order surface in compression space .
however we must be careful : it is possible that all @xmath112-order coefficients @xmath114 ( i.e. those with for all @xmath103 ) are equal to zero , in which case the degree of the surface is strictly less than @xmath47 .
for example , when the gell - mann basis reduces to the pauli basis and we have already seen in the previous section that all eigenvalues are linear in the compression coefficients ; in this case the surface is a plane and not a quadratic surface . in general
then , we may only conclude that is a surface of order at most @xmath47 .
we have now proved that the set of all vectors @xmath30 which induce depolarizing channels @xmath19 ( with respect to the gell - mann basis ) form a finite region in compression space which is bounded by * hyperplanes ( are given by and the remaining one is given by ) * a surface with order at most @xmath115 ( ) * * a surface with order at most @xmath0 ( ) note that the above does not tell us whether the eigenvalues are linear in the compression coefficients .
it turns out that when at least one of the @xmath91 must be non - linear in the compression coefficients a fact which is proved in lemma [ lem : gm - curves ] in section [ sec : other - bases ] . whilst it is a slight disappointment that we have been unable to explicitly give expressions for all the eigenvalues of @xmath86 we can draw some solace from the fact that we have simplified the problem somewhat : in order to see if a vector @xmath30 induces a depolarizing channel we must check to see if all the eigenvalues of the choi - jamiolkowski representation @xmath86 of the induced map @xmath19 are non - negative . without the above results we would have to use a `` brute - force '' algorithm to calculate the @xmath78 eigenvalues of @xmath86 directly ; however the above results enable us to calculate of these quickly , leaving only @xmath0 to be calculated by the brute - force algorithm , which is a substantial speed - up .
in this section we work in the heisenberg - weyl basis which is an alternative generalization of the single - qubit pauli basis to arbitrary dimension .
it has certain advantages over the gell - mann basis ( for example , all the heisenberg - weyl basis matrices are invertible while most of the gell - mann basis matrices are singular ) .
however there is a price to pay for such convenience which is that the heisenberg - weyl basis is not hermitian and so the compression space is a complex vector space .
let us first define @xmath116 where is the primitive @xmath117-root of unity and we work modulo @xmath0 .
note that for @xmath118 and that the inverse of @xmath119 is @xmath120 and similarly for @xmath121 .
we now define the * heisenberg - weyl * basis to be @xmath122 note that this basis is trace - free and trace - orthogonal .
we can now find the eigenvectors and eigenvalues of @xmath86 by defining @xmath123 then it is a simple matter to check that @xmath124 which shows us that @xmath125 are eigenvectors of @xmath86 .
furthermore , there are @xmath78 distinct @xmath125 and we have therefore found all the eigenvectors of @xmath86 . relabeling the compression coefficients @xmath21 as @xmath126 ( with ) we see that the eigenvalues of @xmath86 are @xmath127 it is easy to check that and also . therefore , by the observation in the introduction , . in the preceding sections we have had one ( real ) compression coefficient for each basis matrix . in this section
we have seen that most of the compression coefficients come in complex - conjugate pairs corresponding to two basis matrices which are , up to a complex multiplicative factor , hermitian conjugates of each other .
note that we still have the same number of free compression parameters ( since for each pair of conjugate basis matrices there are two free parameters in the associated compression coefficient , namely the real and imaginary parts ) ; and so the compression space ( although now complex ) still has dimension ( recall that we suppress the dimension corresponding to @xmath22 as it is identically equal to @xmath56 ) .
as an aside , we find the exact structure of the compression space .
we use to see that precisely when and .
when @xmath0 is odd the only solution is and ( i.e. ) . in this case compression space consists complex planes . when @xmath0 is even there are four solutions , namely . in this case ,
compression space consists of @xmath128 real axes and complex planes .
we now prove that , when working in the heisenberg - weyl basis , the set of all compression vectors forms a simplex .
we then find the extremal channels whose compression vectors lie at the vertices of this simplex .
when we work in the heisenberg - weyl basis , the set of all compression vectors forms a simplex in compression space .
it is evident from equation ( [ eqn : lambda - hw ] ) that the eigenvalues of @xmath86 are linear in the compression coefficients and so defines a hyperplane in compression space . since @xmath19 is completely positive if and only if all eigenvalues @xmath129 are non - negative , the hyperplanes must enclose precisely the set of vectors which induce completely positive depolarizing channels @xmath19 .
in particular the hyperplanes enclose a finite region of compression space and the shape of this enclosed region must therefore be a simplex .
the extremal channels are @xmath130 we fix @xmath47 and @xmath103 and prove that @xmath131 is extremal . using the identity
we see @xmath132 and so is a depolarizing channel with compression coefficients .
we now show that there exists @xmath4 and @xmath133 with .
@xmath134 so picking and ensures that . as before we make use of the fact that the eigenvalues of the choi - jamiolkowski representation @xmath135 sum to the system dimension @xmath0 .
clearly @xmath131 is completely positive and so the eigenvalues of @xmath135 are non - negative and sum to @xmath0 . clearly then @xmath136 therefore the compression vector corresponding to the channel @xmath131 lies on all the hyperplanes except one and so it lies at a vertex of the simplex . to finish note that there are @xmath78 vertices and @xmath78 distinct extremal channels , so we have found them all . as before ( when we were working in the pauli - basis ) we note that any compression vector in the simplex can be written as a convex linear combination of the extremal compression vectors ; therefore any depolarizing channel can be written as a convex linear combination of the extremal channels and vice - versa .
having worked in specific bases in the previous sections we now work in an arbitrary trace - free , trace - orthogonal basis @xmath18 .
we have seen that the set of all compression vectors is a simplex precisely when the eigenvalues of @xmath86 are linear combinations of the compression coefficients @xmath21 .
we now find all bases for which this happens .
[ thm : simplex ] let @xmath18 be a trace - free , trace - orthogonal basis .
then the set of all compression vectors forms a simplex if and only if @xmath137 = 0 \quad \forall \alpha,\beta\in\{0,\ldots , n^2 - 1\}\ ] ] the set of all compression vectors forms a simplex if and only if all the eigenvalues of @xmath86 are linear in the compression coefficients ( in which case if @xmath138 is an eigenvalue of @xmath86 then defines a hyperplane which forms one of the sides of the simplex ) .
let @xmath139 be an eigenvector of @xmath86 with eigenvalue @xmath138 .
then by equation ( [ eqn : j(phi ) ] ) we see that that @xmath138 is a linear combination of the compression coefficients @xmath21 if and only if @xmath139 is a simultaneous eigenvector of each ; this occurs if and only if all the are simultaneously diagonalizable , which occurs if and only if for all @xmath23 and @xmath61 . note that the condition of theorem [ thm : simplex ] is weaker than for all @xmath23 and @xmath61 . for example , satisfies @xmath140=0 $ ] but not .
furthermore , it is not hard to show that @xmath141 is equivalent to
. we can now prove the outstanding result from section [ sec : gm - basis ] , which we present as the following lemma .
[ lem : gm - curves ] when and we work in the gell - mann basis then the set of all compression vectors has at least one curved side .
equivalently , at least one of the eigenvalues of @xmath86 is non - linear in the compression coefficients @xmath21 .
it is a simple matter to check that @xmath142 \neq 0\ ] ] then apply theorem [ thm : simplex ] above . for the remainder of this section
we restrict ourselves to working in a basis which is unitary : for all @xmath23 .
when we do this we are able to find the extremal channels whose compression vectors lie at the vertices of the simplex . before proving this
we first define @xmath143 note that @xmath144 is a maximally entangled state and so the @xmath145 are too ( as the @xmath29 are unitary ) .
if for all @xmath23 and @xmath61 and if @xmath29 is unitary for all @xmath23 then @xmath145 are all the eigenstates of @xmath86 .
note that @xmath146 .
note also that unitarity of @xmath29 ensures that .
then @xmath147 and it is easy to see that @xmath148 so @xmath145 is indeed an eigenstate of @xmath86 . finally note that there are @xmath78 eigenstates @xmath145 so we have found all the eigenstates of @xmath86 .
the technical lemma above allows us to find the extremal channels of the simplex : if for all @xmath23 and @xmath61 and if @xmath29 are all unitary then the extremal channels are @xmath149 first note that @xmath150 is completely positive and trace - preserving , so the eigenvalues of @xmath151 are all positive and sum to @xmath0
. we can expand @xmath152 which shows that @xmath153 is a depolarizing channel with compression coefficients .
now observe that @xmath154 and so for all .
clearly then @xmath153 is an extremal channel as its compression vector lies on all the hyperplanes except .
to finish , note that there are @xmath78 vertices to the simplex and there are @xmath78 extremal channels @xmath153 so we have found them all .
we now investigate the effect of changing basis , which will enable us to see why in some bases the set of all compression vectors forms a simplex whilst in other bases it does not .
let and be trace - free , trace - orthogonal bases
. then we may write @xmath155 trace - orthogonality of both bases ensures that the change - of - basis matrix is unitary .
furthermore if both bases are hermitian then @xmath156 is actually a real orthogonal matrix .
if we define @xmath157 and similarly for @xmath158 , then we may write .
we are now in a position to prove [ thm : qubit - simplices ] the set of single qubit ( ) compression vectors forms a simplex whenever we work in a hermitian , trace - free and trace - orthogonal basis . furthermore , the extremal channels are conjugations by the basis matrices . we know that any single qubit basis which is hermitian , trace - free and trace - orthogonal can be obtained from the pauli basis by an orthogonal change of basis .
let be the pauli basis and be the new basis .
then the change of basis matrix has the form @xmath159 where @xmath160 is a orthogonal matrix .
now , any such matrix @xmath160 corresponds to a rotation of the bloch ball and is equivalent to a unitary conjugation where @xmath161 is a unitary matrix .
any such map is completely positive and trace - preserving .
depolarization with respect to the basis is the same as applying the reverse of the above rotation , followed by depolarization in the pauli basis , followed by the the above rotation . since the set of all compression vectors forms a simplex when we work in the pauli basis , the set of all compression vectors also forms a simplex when we work in the new basis .
we know that in the pauli basis the extremal channels are .
so when we start in the new basis and rotate to the pauli basis , the extremal channels will be ; rotating back to the basis ( and noting that ) we see that the extremal channels are . in the single qubit case it is true that any unitary change of basis @xmath156 corresponds to an orthogonal rotation of the polarization vector @xmath14 ;
in higher dimensions this is not so ass the following example demonstrates [ ex : bloch - rotation ] let and let @xmath162 where is the single - qubit pauli-@xmath163 matrix .
one can explicitly calculate the eigenvalues of @xmath3 ; they are each with multiplicity two , so @xmath3 is a state when . let us now apply any change of basis which maps ( that is , we change basis from the hermitian pauli basis to the hermitian gell - mann basis with a unitary change - of - basis matrix @xmath156 ) .
then @xmath164 which has eigenvalues ( with multiplicity 3 ) and ( with multiplicity 1 ) .
but then if then @xmath3 is a quantum state but @xmath165 is not ( as it is not positive ) .
this example demonstrates that some changes of basis do not map the bloch `` ball '' on to itself .
it is for this reason that the set of all compression vectors forms a simplex in some bases but not in others . when two trace - free , trace - orthogonal bases and are related via @xmath166 for all @xmath23 and some unitary matrix @xmath161 _ ( that is , the change - of - basis matrix @xmath156 maps the bloch `` ball '' on to itself ) _ ,
then the set of compression vectors either forms a simplex in both bases or does not form a simplex in either basis . furthermore , if the extremal channels are conjugations of basis matrices in one basis then they are conjugations of basis matrices in the other basis too .
one may adapt the proof of theorem [ thm : qubit - simplices ] .
( note that the condition for all @xmath23 guarantees that the `` bloch ball '' is rotated onto itself and we avoid the problems exhibited in example [ ex : bloch - rotation ] . )
throughout the previous sections of this work we have studied depolarizing channels .
our method can be extended to deal with a more general class of channels , namely those which both compress the bloch `` ball '' and then translate it .
let @xmath167 be such a channel defined by its action on the polarization vector @xmath168 we call the vector the * translation vector * and the @xmath169 the * translation coefficients*. note that to ensure that @xmath167 is trace - preserving .
so far we have only studied channels , but we now extend to general @xmath170
. we may expand @xmath171 if we now define @xmath172 for all @xmath23 ; and @xmath173 for and then @xmath174 recalling the fact that if a channel @xmath175 has the expression then the choi - jamiolkowski representation is we see that @xmath176 which enables us to find the choi - jamiolkowski representation of the channel @xmath167 @xmath177 in order to find which parameters @xmath178 induce completely positive channels one has to find for which parameters all the eigenvalues of @xmath179 are non - negative .
( recall that these channels are automatically trace - preserving if and ) .
let us work in the pauli basis and consider a single - qubit channel @xmath167 .
we fix the translation vector and find the set of all compression vectors @xmath30 .
now , @xmath180 it turns out that for non - zero @xmath181 the eigenvectors of @xmath179 are no longer bell - states , but rather linear combinations of them .
let @xmath182 where are real numbers , and @xmath183 ( ) are the bell - states defined in section [ sec : p - basis ] .
one can easily check that @xmath184 where the eigenvalues satisfy the following relations @xmath185 which can be solved to give @xmath186 note that there are two possible values for @xmath187 and @xmath188 : one for and one for ; we have therefore found all four eigenvalues of @xmath179 . clearly for the surfaces and are curved .
this reproduces the results of @xcite , which also shows that the set of all compression vectors forms an `` asymmetrically rounded tetrahedron '' .
we could have worked through the above example with @xmath189 or @xmath190 non - zero but then the generic form of the eigenvectors of @xmath179 would be more complicated : with the coefficients satisfying .
based on the above example we note that even when we work in a basis where the set of all compression vectors forms a simplex for , a tiny perturbation to destroys this simplex : the set of compression vectors @xmath30 forms a set with a curved boundary ; in this situation there are infinitely many extremal channels .
in summary , we defined _ trace - free , trace - orthogonal bases _ for quantum states and used these to define _ depolarizing channels_. each depolarizing channel has an associated _ compression vector _ which lies in _ compression space _ which is an dimensional vector space ( we suppressed the other dimension as the first component of the compression vector is always equal to unity ) .
we showed that the set of all compression vectors forms a simplex when we work in either the _ pauli basis _ or the _ heisenberg - weyl basis _ , but that it does not form a simplex when we work in the _ gell - mann _ basis .
furthermore , in the pauli and heisenberg - weyl bases we found the extremal channels whose compression vectors lie at the vertices of the corresponding simplex . working in a general trace - free , trace - orthogonal basis
, we showed precisely for which bases the set of compression vectors forms a simplex .
when the basis matrices were all unitary we were able to find the extremal channels .
we discussed the effects of changing basis and this indicated why the set of compression vectors forms a simplex in some bases but not in others .
finally we generalized our methods to deal with a class of more general quantum channels .
i would now like to recall a theorem proved in @xcite but first i must define a * doubly stochastic * quantum channel acting on an @xmath0-dimensional quantum system to be any completely - positive , trace - preserving channel which leaves the completely mixed state untouched : .
let us now define @xmath191 to be the set of all doubly stochastic channels which act on an @xmath0-dimensional quantum system .
as @xmath191 is a convex set we may define @xmath192 to be the set of extremal channels in @xmath191 . it is known that all doubly stochastic single - qubit channels are mixed unitary , but that in higher dimensions ( ) there are doubly stochastic channels which are not mixed unitary .
( they can however be expressed as an affine sum of unitary channels where are affine parameters : and . ) in particular , when and we work in any unitary basis for which the set of compression vectors forms a simplex , there exist doubly stochastic channels which do not lie within the simplex and are therefore not depolarizing channels .
however , the following theorem ( proved in @xcite ) tells us that all such channels when appropriately averaged with the isotropic completely depolarizing channel @xmath194 , become mixed unitary .
let @xmath175 be any doubly - stochastic quantum channel acting on @xmath0-dimensional quantum systems and let @xmath33 be any depolarizing channel whose compression vector lies within a simplex whose vertices are the compression vectors of unitary conjugation depolarizing channels .
then there exists a constant such that for the following channel is mixed unitary @xmath197 | a generalized depolarizing channel acts on an @xmath0-dimensional quantum system to compress the `` bloch ball '' in directions ; it has a corresponding compression vector .
we investigate the geometry of these compression vectors and prove a conjecture of dixit and sudarshan @xcite , namely that when ( i.e. the system consists of @xmath1 qubits ) and we work in the pauli basis then the set of all compression vectors forms a simplex .
we extend this result by investigating the geometry in other bases ; in particular we find precisely when the set of all compression vectors forms a simplex . |
evidences for the dark matter ( dm ) are provided by many astrophysical observations , although the nature of dm is yet to be uncovered .
the most popular candidates for dm are weak interacting massive particles ( wimp ) , which we shall assume in this work .
dm can be detected either directly or indirectly with the former observing the nucleus recoil as dm interacts with the target nuclei in the detector and the latter detecting final state particles resulting from dm annihilations or decays .
the direct detection is possible because that the dark matter particles constantly bombard the earth as the earth sweeps through the local halos .
as just stated , the direct detection experiments record the nuclei recoil energy of nuclei - wimps scattering . at present
, dama @xcite , cogent @xcite , and cresst @xcite have reported the detection of dm signal with the dm mass @xmath7 ranging from few gev to 50 gev and the spin - independent scattering cross section @xmath8 pb . on the other hand
, xenon100 @xcite only collects 2 events which are consist with the background .
this result then sets the limit @xmath9 pb for @xmath10 gev .
interestingly , recent cdms ii result @xcite reports three signal events which gives a p - value 0.19% ( less than @xmath11 ) .
the corresponding best - fit values of dm parameters are @xmath12 gev and @xmath13 pb . dm can also be detected indirectly by measuring the positron signals from milky way .
pamela observed a rise of the cosmic ray positron fraction for positron energy greater than 10 gev @xcite .
this anomalous enhancement are confirmed by fermi - lat @xcite and the recently released ams02 first result @xcite . in the recent ams02 result ,
this continuous rise on positron fraction is extended up to positron energy @xmath14 350 gev .
such a spectral behavior makes the dm annihilation explanation of the data difficult because it requires a large boost factor for annihilation cross section @xmath15 provided the thermal equilibrium for dm in the early universe is reached with @xmath16 .
for example , several groups @xcite fit the updated galactic positron fraction with ams02 new result included .
they have found that the favored dm parameter region is located at @xmath17 few tev and @xmath18 if dm annihilation channel @xmath19 is responsible for the positron excess .
the favored dm mass range can be lowered to few hundred gev if nearly pulsar sources are considered together with @xmath19 annihilations @xcite .
however the favored values for @xmath15 do not decrease much in such a combined fitting .
it is important to note that dm annihilations in the galactic halo are constrained by fermi - lat gamma ray observations @xcite .
the constraint on @xmath20 is in fact located in the preferred dm parameter region resulting from pamela and ams02 measurements for the same range of @xmath4 .
it has been pointed out some time ago that the preferred dm parameter region by pamela and fermi - lat measurements can be examined through the observation of neutrinos @xcite(see also discussions in refs .
@xcite ) by icecube detector augmented with deepcore array .
indeed icecube 22 string result on searching for dm annihilations from galactic halo @xcite has set the upper limit for @xmath20 comparable to the required annihilation cross section for explaining pamela and fermi - lat data .
the icecube sensitivity on dm signature from the galactic halo is expected to improve with the data from all 86 strings analyzed .
the analysis of deepcore array data will further enhance the sensitivity in the small @xmath7 regime @xcite which is of interest due to direct detection results mentioned above .
it is interesting to note that the constraints on dm capture cross section and annihilation cross section @xmath15 can be obtained from the searching for dm annihilations from the earth core .
the detection of dm induced neutrino signature from the earth core has been discussed previously @xcite .
it has been shown that the chemical composition of the earth core results in several dm annihilation peaks for @xmath4 ranging from 20 gev to 60 gev .
these peaks do not appear for annihilations inside the sun .
furthermore , dm annihilation rate inside the sun is completely determined by the capture cross section ( dominated by spin dependent component proportional to @xmath21 ) while dm annihilation rate in the earth core depends on both @xmath6 ( contribution proportional to @xmath21 is negligible ) and @xmath15 .
this is understood by the fact that dm density in the former case has already reached equilibrium while dm density in the latter case has not .
hence the search for neutrino signature from the earth core can probe both cross sections .
model - independent sensitivity studies on icecube detection of dm induced neutrino signature from the earth core were reported in @xcite for dm mass around tev . in this work
, we consider an extended dm mass range from few tens of gev to tev . in the low mass range ,
our results can be compared with direct detection results from dama , cogent , cresst - ii and xenon100 mentioned above . in the high mass range around tev , our results can be compared with those from cosmic ray observations by pamela , fermi - lat and ams02 .
we study both muon track events and cascade events induced by neutrinos .
we consider annihilation channels @xmath22 for signature neutrino productions . for non - monochromatic modes ,
we note that @xmath23 produces hardest neutrino spectrum while @xmath24 produces the softest one .
we also include @xmath25 oscillations for lower energy neutrinos .
this paper is organized as follows . in sec .
ii , we discuss the neutrino flux produced in the earth core by dm annihilations .
the procedure for calculating such a flux is outlined . in sec .
iii , we discuss the track and shower event rates resulting from dm annihilations in the earth core .
the background event rates from atmospheric neutrino flux are also calculated .
we adopt the effective areas published by icecube observatory for event rate calculations . in section iv , we present icecube / deepcore 5-year sensitivities for detecting dm induced neutrino signature from the earth core .
we first fix the dm mass at two representative values , @xmath26 gev and @xmath27 tev . the icecube / deepcore @xmath28 sensitivity for 5-year data taking
is then presented as a curve on @xmath29 plane .
next , we fix the annihilation cross section @xmath15 at conservative values , @xmath30 and @xmath31 .
we then present icecube / deepcore sensitivities to spin - independent cross section @xmath6 as a function of @xmath4 for different assumptions on dominant dm annihilation channels .
such sensitivities are then compared with existing constraints from direct detection experiments and that obtained from the icecube / deepcore search of dm annihilations in the sun . finally , we take different experimental bounds on @xmath6 as inputs to obtain different icecube / deepcore sensitivities to dm annihilation cross section on the @xmath32 plane for different annihilation channels .
there are thus three scenarios for the input @xmath6 : ( a ) @xmath6 favored by dama and cresst - ii ; ( b ) @xmath6 bound set by xenon100 ; ( c ) @xmath6 bound set by xenon1 t ( 2017 ) @xcite assuming non - detection .
we particularly compare icecube sensitivity to @xmath20 to the favored range on the same quantity implied by pamela , fermi - lat and ams02 .
we note that neither @xmath33 nor @xmath24 can simultaneously fit well to pamela and ams02 @xcite if either channel is assumed to be dominant .
we conclude in sec .
to facilitate our discussions , let us define @xmath34 as the energy spectrum of @xmath35 produced per dm annihilation @xmath36 in the earth core .
the differential dm neutrino flux of flavor @xmath37 on the earth surface is then given by @xmath38 where @xmath39 is the dm annihilation rate , @xmath40 is the branching ratio for the dm annihilation channel @xmath36 , @xmath41 is the distance between the source and the detector , and @xmath42 is the neutrino oscillation probability from the source to the detector . to calculate @xmath43 ,
we employ ` wimpsim ` @xcite with a total of 50000 monte carlo generated events .
although we are particularly interested in icecube / deepcore measurements , the dm neutrino flux from the earth core is the same for all detector locations near the earth surface due to the spherical symmetry .
the oscillation probability @xmath42 is calculated with the best fit neutrino oscillation parameters summarized in table i of ref .
@xcite , @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 @xmath49 , and @xmath50 @xmath49 .
the dm annihilation rate @xmath39 can be determined by the following argument .
when the earth sweeps through dm halo , the wimp could collide with matter inside the earth and lose its speed .
if the wimp speed becomes less than its escape velocity , the wimp can be captured by earth s gravitational force and then sinks into the core of earth . after a long peroid of accumulation
, wimps inside the core of earth can begin to annihilate into standard model particles at an appreciable rate . among the annihilation final states , neutrino can be detected by neutrino telescopes .
let @xmath51 be the number of wimps in the earth core at time @xmath52 , we have @xmath53 where @xmath54 is the capture rate and @xmath55 is the evaporation rate .
it has been shown that wimps with masses between @xmath56 gev may evaporate from the earth @xcite
. since we are interested in the mass range @xmath57 gev , we neglect @xmath55 in our discussions .
the capture rate @xmath54 depends on dm - nuclei elastic scattering cross section which contains spin - dependent component @xmath21 and spin - independent component @xmath6 .
the dm annihilation rate @xmath58 is proportional to @xmath59 .
one writes @xmath60 taking into account the quasi - thermal distribution of wimps in the earth core , the annihilation coefficient @xmath61 can be written as @xcite @xmath62 where @xmath63 @xmath64 for the earth . by solving @xmath51 in eq .
( [ eq : x2 ] ) , we obtain @xcite @xmath65 where @xmath52 is the age of the macroscopic body , for example @xmath66 gyr for sun and earth , while @xmath67 is the equilibration time scale , @xmath68 .
numerically @xmath69 for @xmath70 .
in such a case @xmath71 so that the dm annihilation rate in the earth core depends only on the capture rate @xmath54 and is independent of the annihilation cross section @xmath15
. since the heavy nuclei such as iron is abundant in the earth core , the capture cross - section is enhanced due to its quadratic dependence on the nuclei atomic number .
therefore , the corresponding capture rate is given by @xcite @xmath72 with @xmath73 the dm velocity dispersion , @xmath74 the local dm density , and @xmath75 the atomic number of chemical element in the earth core .
@xmath76 is the product of various factors including the mass fraction of element @xmath75 , the gravitational potential for element @xmath75 , kinematic suppression factor , form factor , and a factor of reduced mass .
the explicit form of @xmath76 is not essential for our discussions .
it can be found , for instance , in ref .
@xcite . however , we like to stress that there are significant astrophysical uncertainties on the dm local density and its velocity distributions involved in @xmath54 @xcite . in this work ,
we use approximate formulae given in @xcite , which are adopted by ` darksusy ` @xcite .
neutrino telescope such as icecube detects neutrinos by measuring muon track and cascade events , which are induced by neutrino - nucleon charged - current ( cc ) and neutral - current ( nc ) scatterings .
we calculate neutrino event rate according to icecube published neutrino effective area @xmath77 @xcite of the full icecube 86-string detector , where @xmath78 is interaction type for different neutrino flavor @xmath37 . for example
, for muon track events , @xmath37 is ( anti- ) muon neutrino and @xmath78 is ( anti- ) muon neutrino cc interaction . on the other hand , for cascade events
, @xmath37 includes all three neutrino flavors . for ( anti- ) electron neutrinos and ( anti- ) tau neutrinos
, @xmath78 runs over cc and nc interactions while @xmath78 is exclusively nc interaction for ( anti- ) muon neutrinos .
the effective area accounts for the detection efficiency including the neutrino - nucleon interaction probability , the energy loss of muon from its production point to the detector , and the detector trigger , and analysis efficiency .
hence , the neutrino event rate from the earth dm is given by @xmath79 where @xmath80 is the differential neutrino flux in the vicinity of detector for a given neutrino flavor @xmath37 , which is given by eq .
( [ eq : dnde ] ) , and the index @xmath78 can be either nc or cc interaction .
the effective area can be defined as @xmath81 @xcite , where @xmath82 @xmath83 is the density of ice , @xmath84 @xmath85 is the avogadro number , @xmath86 is the neutrino - nucleon cross - section , and @xmath87 is the effective volume of icecube for different neutrino - nucleon interaction events . in this work , we shall take the detector threshold energy @xmath88 for icecube / deepcore as 10 gev and 100 gev , respectively . to compute the rate for @xmath89 track events
, we shall use the effective area ( deepcore+icecube trigger ) given by ref .
@xcite .
unlike the calculation of track event rate , which requires only one effective area , the calculation of cascade event rate requires 5 different effective areas . the cascade - event effective area given by ref .
@xcite is only for @xmath90 cc interaction events .
here we also adopt the effective area marked as deepcore+icecube trigger for estimating the @xmath90 event rate . to obtain effective areas for other cascade events , we perform the rescaling @xmath91 where @xmath92 is the averaged fraction of neutrino energy @xmath93 converted into shower energy after a neutrino - nucleon cc or nc interaction .
one has @xmath94 for nc interactions of @xmath90 and @xmath89 while @xmath95 for @xmath90 cc interaction @xcite . for @xmath96 ,
the fraction @xmath92 resulting from @xmath97 cc interaction and subsequent tau - lepton decay is approximately @xmath98 where @xmath99 is the energy fraction of @xmath97 taken by hadrons in @xmath97-nucleon cc interaction @xcite .
the factor 0.4 is the visible energy fraction in tau lepton decays , which can be estimated by using pythia @xcite .
it should be noted that the final effective area / effective volume for cascade events may be significantly reduced due to the background rejection cut . in a recent icecube 79-string result on atmospheric @xmath90 flux measurement @xcite ,
the final effective volume for @xmath90 events is much smaller than the one at the deepcore filter level for lower energies while such a difference is within an order of magnitude for @xmath100 gev , as can be seen from fig . 2 of that paper .
since our sensitivity calculations for cascade events are based upon effective areas in ref .
@xcite , we only present icecube / deepcore sensitivities for @xmath101 gev in the case of cascade events . for calculating the atmospheric ( atm ) background event rates
, we use the atmospheric neutrino flux taken from ref . @xcite . because the intrinsic atm @xmath96 flux is negligible ,
we only consider atm @xmath90 and @xmath89 fluxes at the production point .
it is essential to include the effect of neutrino transmissions through the earth .
since the earth becomes opaque to neutrinos only for @xmath102 tev @xcite , the neutrino transmission in our interested energy range is essentially the neutrino oscillation effect .
hence the atm background event rate is given by @xmath103 where @xmath104 is atm neutrino flux in the vicinity of the detector .
such a flux is given by @xmath105 where @xmath106 is the atm neutrino flux at the source , @xmath107 is the neutrino oscillation probability with @xmath108 the neutrino traversing distance through the earth along the direction of @xmath109 . for comparing with the signal event rate induced by earth dm annihilation with dm mass @xmath4 ,
@xmath110 is taken as @xmath4 in eq .
( [ eq : neventb1 ] ) .
we remark that @xmath110 should differ from @xmath7 in practice due to the energy resolution effect .
however the impact of energy resolution on our sensitivity estimation will be shown insignificant in the next section .
( left ) and @xmath4 ( right ) .
, title="fig:",scaledwidth=49.0% ] ( left ) and @xmath4 ( right ) . , title="fig:",scaledwidth=49.0% ] in fig .
[ fig : nevents ] , we show the number of neutrino events in 5 years from dm annihilations ( @xmath23 ) and atm background as functions of the maximum open angle @xmath111 ( left ) and the dm mass @xmath4 ( right ) , respectively . here , we take @xmath112 and @xmath6 the 90% cl upper limit from xenon100 . from eq .
( [ eq : ca ] ) , we can see the annihilation coefficient @xmath61 is inverse proportional to a @xmath4 dependent effective volume @xmath113 , which describes the volume of dark matter occupation in the earth core .
hence it is seen from the left panel of fig .
[ fig : nevents ] that @xmath114 reaches to the maximum for sufficiently large @xmath111 that can cover the entire dm populated region in the earth core .
the critical value of @xmath111 for covering the dm populated region in the earth core is a function of @xmath4 , which we denote as @xmath115 .
we have @xmath116 , \label{eq : phimax}\ ] ] where @xmath117 the radius of the earth .
the @xmath118 on the right hand side of the equation is to ensure a minimal open angle of @xmath118 .
we have seen that @xmath119 decreases as @xmath4 increases .
hence , @xmath120 for @xmath121 gev is smaller than @xmath120 for @xmath26 gev , as can be seen from the left panel of fig .
[ fig : nevents ] . the right panel of fig .
[ fig : nevents ] shows dm and atm background event numbers as functions of @xmath4 where @xmath111 for each @xmath4 is taken to be @xmath122 .
the @xmath123 for dm signal peaks at three different values of @xmath4 .
this is due to the enhancement of capture rate when @xmath4 is close to the mass of any dominantly populated nuclei in the earth core .
in fact , the three peaks from small to large @xmath4 correspond to the resonant capture by oxygen , mg / si , and fe / ni , respectively . effects of these resonant capture peaks have been studied in inert doublet @xcite and supersymmetry neutralino dm models @xcite . in ref .
@xcite , the authors also included the effect from dark disc and found that the search for dm annihilation in the earth can have the same level of sensitivity as the search for dm annihilation in the sun for @xmath124 gev .
we present the sensitivity as a @xmath28 detection significance in 5 years , calculated with the standard formula @xmath125 the atm here is the number of atmospheric background events , which we calculate with the flux data from ref . @xcite . the right hand side of eq .
( [ eq : sensitivity ] ) refers to the 2@xmath126 detection significance .
as mentioned earlier , the effect of detector energy resolution on the sensitivity estimation needs to be understood .
here we assume icecube can determine the neutrino energy with the energy resolution @xmath127 @xcite . with this energy resolution
, we may compare the atm track event rate in eq .
( [ eq : neventb1 ] ) with @xmath128 and the similar event rate with @xmath129 , which has taken into account the @xmath130 energy resolution .
we have found that , for @xmath131 gev and @xmath132 , the ratio of the atm background event rate with @xmath129 to that with @xmath128 varies between 2.7 and unity . from eq .
( [ eq : sensitivity ] ) , we can see that the magnitude of dm annihilation cross section which the detector can probe is proportional to the square root of the atm background event number .
thus the icecube sensitivity to @xmath2 with track events is changed only slightly by a factor @xmath133 in the range @xmath134 .
we expect similar effect for cascade events . therefore , the effect of energy resolution on our sensitivity estimation is insignificant
. significance on ( @xmath15 , @xmath6 ) plane . in the left panel ,
the dm mass is 50 gev and the threshold energy is 10 gev .
the dm mass is 2 tev and threshold energy is taken to be 100 gev in the right panel .
solid lines are sensitivities with track events while dashed lines are sensitivities with cascade events . the pink shaded area represents the 90% exclusion from xenon100 ( 2012 ) result .
, title="fig:",scaledwidth=49.0% ] significance on ( @xmath15 , @xmath6 ) plane . in the left panel ,
the dm mass is 50 gev and the threshold energy is 10 gev .
the dm mass is 2 tev and threshold energy is taken to be 100 gev in the right panel .
solid lines are sensitivities with track events while dashed lines are sensitivities with cascade events .
the pink shaded area represents the 90% exclusion from xenon100 ( 2012 ) result .
, title="fig:",scaledwidth=49.0% ] in fig .
[ fig : sigsip_bf ] , we show the 5 year sensitivity of the full icecube 86-string detector to earth dm signal on ( @xmath15 , @xmath6 ) plane .
we consider annihilation channels @xmath135 .
the first annihilation mode produces cascade events while the second and third channels produce both track and cascade events due to @xmath25 oscillations .
each annihilation channel is assumed to be dominant when icecube / deepcore sensitivity to that channel is derived . the open angle @xmath136 for collecting events from the earth core is taken to be @xmath137 for both track and cascade events .
this is reasonable for track events and achievable for high energy cascade events @xcite such as those produced with @xmath27 tev described in the right panel .
since we only consider cascade events with @xmath138 , the sensitivities with cascade events are only presented in the right panel . in the right panel
, we can see that @xmath139 channel is most sensitive to @xmath6 .
in particular , this channel provides better sensitivity than @xmath140 channel due to relatively less atm background events .
we note that the 5 year sensitivity curve on ( @xmath15 , @xmath6 ) plane is almost linear in logarithmic scale so that @xmath6 approximates to @xmath141 with slope @xmath142 .
as @xmath15 increases , a smaller scattering cross section @xmath6 is sufficient to achieve the same detection significance .
however , eq . ( [ eq : x1 ] ) implies that this trend can not continue indefinitely .
as @xmath143 is driven to the plateau by a sufficiently large @xmath15 , @xmath54 must approach to a constant value for maintaining the same annihilation rate @xmath39 .
this then implies that @xmath6 also approaches to a constant value . in the reverse direction where @xmath15 decreases ,
a larger @xmath6 is required to achieve the same detection significance . on the other hand ,
the xenon100 limit ( pink shaded region ) eventually sets the upper bound for @xmath6 .
this constraint is clearly seen for @xmath23 channel with @xmath144 tev . due to xenon100 limit
, such a channel can not produce enough neutrino events in icecube for reaching @xmath28 sensitivity in 5 years , unless the boost factor for @xmath15 is larger than @xmath145 .
it is clear that xenon100 limit sets a @xmath7-dependent bound on @xmath15 for each annihilation channel .
however , if one takes different experimental bounds on @xmath6 , the prospect of observing neutrinos from the earth core differs drastically .
there are thus three possibilities : case a : neutrino observation implied by dama and cresst - ii favored parameter space ; + case b : @xmath15 exclusion limits implied by xenon100 bound on @xmath6 ; + case c : pessimistic scenario by assuming non - detection of xenon1 t ( 2017 ) . below we shall discuss in turn these three cases . ,
@xmath6 ) plane for @xmath146 , and @xmath147 annihilation channels .
the upper figures are based on @xmath148 while the lower figures are based on @xmath149 .
the track - event sensitivities with @xmath150 are presented in figures on the left column and that with @xmath151 are presented in figures on the right column .
the energy threshold is taken to be 10 gev .
the icecube 79-string upper limits on @xmath152 ( red dashed - squared line ) and @xmath153 ( red solid - squared line ) from the search for dm - induced neutrino signature from the sun are also shown for comparison @xcite .
, title="fig:",scaledwidth=49.0% ] , @xmath6 ) plane for @xmath146 , and @xmath147 annihilation channels .
the upper figures are based on @xmath148 while the lower figures are based on @xmath149 .
the track - event sensitivities with @xmath150 are presented in figures on the left column and that with @xmath151 are presented in figures on the right column .
the energy threshold is taken to be 10 gev .
the icecube 79-string upper limits on @xmath152 ( red dashed - squared line ) and @xmath153 ( red solid - squared line ) from the search for dm - induced neutrino signature from the sun are also shown for comparison @xcite .
, title="fig:",scaledwidth=49.0% ] + , @xmath6 ) plane for @xmath146 , and @xmath147 annihilation channels .
the upper figures are based on @xmath148 while the lower figures are based on @xmath149 .
the track - event sensitivities with @xmath150 are presented in figures on the left column and that with @xmath151 are presented in figures on the right column .
the energy threshold is taken to be 10 gev .
the icecube 79-string upper limits on @xmath152 ( red dashed - squared line ) and @xmath153 ( red solid - squared line ) from the search for dm - induced neutrino signature from the sun are also shown for comparison @xcite .
, title="fig:",scaledwidth=49.0% ] , @xmath6 ) plane for @xmath146 , and @xmath147 annihilation channels .
the upper figures are based on @xmath148 while the lower figures are based on @xmath149 .
the track - event sensitivities with @xmath150 are presented in figures on the left column and that with @xmath151 are presented in figures on the right column .
the energy threshold is taken to be 10 gev .
the icecube 79-string upper limits on @xmath152 ( red dashed - squared line ) and @xmath153 ( red solid - squared line ) from the search for dm - induced neutrino signature from the sun are also shown for comparison @xcite .
, title="fig:",scaledwidth=49.0% ] + in fig .
[ fig : mx_sigsip ] , we present icecube / deepcore sensitivities to @xmath154 , and @xmath147 annihilations in the earth core on the ( @xmath7 , @xmath6 ) plane for track events with different @xmath111 and @xmath15 with @xmath131 gev .
the sensitivities to @xmath155 ( black dashed - doted ) and @xmath23 ( grey solid ) are comparable , albeit the @xmath1 channel only opens at @xmath156 .
the experimental upper limits of @xmath152 ( red dashed - squared line ) and @xmath157 ( red solid - squared line ) are taken from icecube 79-string result on the search for muon neutrino events induced by dm annihilations in the sun @xcite .
we note that the constraint on @xmath158 stands for a constraint on @xmath155 for @xmath159 and a constraint on @xmath23 for @xmath160 . for @xmath124
gev , it is seen that @xmath6 can be better probed by detecting dm induced neutrino signature from the earth core than that from the sun .
the favoured region of dama at higher @xmath7 is not compatible with icecube 79-string constraint on @xmath23 .
this region can also be probed by searching for neutrinos from @xmath23 and @xmath147 annihilations in the earth core .
in fact , the search for muon track events induced by @xmath23 with @xmath161 can probe the full allowed region of dama and most of the allowed region of cresst - ii . with a 10 times smaller @xmath15 , the full allowed region of dama can still be probed by the same annihilation channel .
significance on the ( @xmath7,@xmath15 ) plane for @xmath162 and @xmath163 .
the @xmath6 input values are taken from no channeling dama and cresst - ii shown in fig .
[ fig : mx_sigsip ] .
the figure in the left panel is for @xmath152 channel while the figure on the right panel is for @xmath23 channel .
the dashed / solid lines correspond to sensitivities obtained with input @xmath6 given by the lower / upper boundary of dama and cresst contours .
the allowed regions for @xmath15 are indicated by hatched areas .
, title="fig:",scaledwidth=49.0% ] significance on the ( @xmath7,@xmath15 ) plane for @xmath162 and @xmath163 .
the @xmath6 input values are taken from no channeling dama and cresst - ii shown in fig .
[ fig : mx_sigsip ] .
the figure in the left panel is for @xmath152 channel while the figure on the right panel is for @xmath23 channel .
the dashed / solid lines correspond to sensitivities obtained with input @xmath6 given by the lower / upper boundary of dama and cresst contours .
the allowed regions for @xmath15 are indicated by hatched areas .
, title="fig:",scaledwidth=49.0% ] from fig .
[ fig : sigsip_bf ] , we can see that the sensitivity is correlated as @xmath164 .
therefore , if one takes dama and cresst - ii favored regions as input , it is possible to probe @xmath15 to a value much smaller than @xmath165 as shown by fig .
[ fig : mx_sigv_dama ] . in this figure
, it is assumed that dm parameter regions are those given by dama and cresst - ii .
we then present the icecube / deepcore 5 year track - event sensitivity curves on the ( @xmath7 , @xmath15 ) plane .
we take the boundary of the discovery contour of dama and cresst - ii as our @xmath6 inputs .
the upper icecube / deepcore sensitivity curves ( thin lines ) are driven by the lower boundary of @xmath6 contours of dama and cresst - ii , while the lower sensitivity curves ( thick lines ) are driven by the upper boundary of @xmath6 contours .
we do not use the data of cogent because their favoured dm mass range is below the energy threshold @xmath166 gev .
for the same reason , we also ignore the @xmath167 gev favoured regions of dama and .
we choose the open angle @xmath163 . because of the large capture rate @xmath54 resulted from the iron resonance region , @xmath15 for this @xmath4 range can be probed to values much smaller than the thermal average cross section @xmath168 @xmath169 .
we can take the xenon100 90% upper limit as the input @xmath6 .
let us begin by taking @xmath131 gev and consider only track events .
the 5-year icecube / deepcore @xmath28 sensitivities to @xmath170 and @xmath171 as functions of @xmath4 are presented in fig .
[ fig : mx_sigmav1 ] .
we note that the annihilation channel @xmath23 can give rise to track events due to the @xmath172 oscillations for lower @xmath173 .
the results are obtained by considering track events with @xmath150 .
the yellow shaded region corresponds to the steady state with @xmath174 , which is caused by a sufficiently large @xmath15 when @xmath6 is fixed at the current xenon100 upper limit . in this case
, the number of dm trapped in the earth core reaches to the equilibrium value since @xmath175 . as a result
, the annihilation rate which dictates the neutrino flux is determined entirely by the capture rate and is independent of @xmath15 .
the latter only determines the number of dm in the equilibrium .
hence the measurement of neutrino flux in the steady state can only determine @xmath6 .
sensitivities to @xmath170 and @xmath171 as functions of @xmath4 with track events .
the @xmath6 is taken from xenon100 90% upper limit .
the yellow shaded area corresponds to the steady state with @xmath174 .
we take @xmath131 gev and @xmath150 .
, scaledwidth=55.0% ] we can see the strongest limit comes from the iron resonance region @xmath176 gev where the capture rate @xmath54 peaks . on the other hand ,
the weakest bound of @xmath15 occurs at the lowest point of xenon100 @xmath6 upper limit located at @xmath177 gev .
moreover , the curves are broken near the boundary of the yellowed region . hence , for each annihilation channel , there exists a range of @xmath7 where the @xmath28 significance curve disappears .
this means that the dm event number in this @xmath7 range can not reach the @xmath28 significance before @xmath15 reaches to the steady state @xmath174 .
one should bear in mind that this yellowed region varies with the input @xmath6 .
sensitivities to @xmath15 of various channels as functions of @xmath4 with track and cascade events .
the threshold energy is taken to be @xmath178 gev and the @xmath7 range is extended to 10 tev .
the cyan contours refer to @xmath179 confidence region of pamela and ams02 combined analysis , for @xmath23 channel , taken from ref .
the black solid line is the 5 year sensitivity upper limit of @xmath180 in @xmath28 significance obtained from the search of neutrino cascade events from the galactic center with @xmath181 relative to the direction of galactic center .
, title="fig:",scaledwidth=49.0% ] sensitivities to @xmath15 of various channels as functions of @xmath4 with track and cascade events .
the threshold energy is taken to be @xmath178 gev and the @xmath7 range is extended to 10 tev .
the cyan contours refer to @xmath179 confidence region of pamela and ams02 combined analysis , for @xmath23 channel , taken from ref .
the black solid line is the 5 year sensitivity upper limit of @xmath180 in @xmath28 significance obtained from the search of neutrino cascade events from the galactic center with @xmath181 relative to the direction of galactic center .
, title="fig:",scaledwidth=49.0% ] + sensitivities to @xmath15 of various channels as functions of @xmath4 with track and cascade events .
the threshold energy is taken to be @xmath178 gev and the @xmath7 range is extended to 10 tev .
the cyan contours refer to @xmath179 confidence region of pamela and ams02 combined analysis , for @xmath23 channel , taken from ref .
the black solid line is the 5 year sensitivity upper limit of @xmath180 in @xmath28 significance obtained from the search of neutrino cascade events from the galactic center with @xmath181 relative to the direction of galactic center .
, title="fig:",scaledwidth=49.0% ] sensitivities to @xmath15 of various channels as functions of @xmath4 with track and cascade events .
the threshold energy is taken to be @xmath178 gev and the @xmath7 range is extended to 10 tev .
the cyan contours refer to @xmath179 confidence region of pamela and ams02 combined analysis , for @xmath23 channel , taken from ref .
the black solid line is the 5 year sensitivity upper limit of @xmath180 in @xmath28 significance obtained from the search of neutrino cascade events from the galactic center with @xmath181 relative to the direction of galactic center . , title="fig:",scaledwidth=49.0% ] + we next take @xmath101 gev and consider both track and cascade events .
we present in fig .
[ fig : mx_sigmav2 ] the icecube / deepcore sensitivities to @xmath15 of various channels as functions of @xmath4 .
the result in the upper left panel is obtained by taking @xmath182 , while @xmath111 is taken to be @xmath137 , @xmath183 and @xmath184 for results in the upper right , lower left and lower right panels , respectively . for @xmath23 channel
, we also plot the @xmath179 confidence region favoured by pamela and ams02 positron fraction data @xcite for comparison .
one can see that the search for @xmath23 cascade events can probe almost all the @xmath179 favoured region by pamela and ams02 with @xmath150 . for comparison , we also estimate the icecube / deepcore sensitivity to the dm annihilation cross section in the galactic halo with a 100 gev threshold energy by using the method of ref .
@xcite and the energy dependent effective volume @xmath185 @xcite .
the 5 year sensitivity upper limit of @xmath180 obtained from searching for neutrino cascade events from the galactic center with @xmath181 relative to the direction of galactic center is also plotted .
we note that this sensitivity is independent of @xmath6 , unlike the search for dm annihilations in the earth core .
one can see that the search for galactic dm annihilations by icecube / deepcore can probe the entire @xmath179 confidence region favoured by pamela and ams02 in 5 years of running .
it is interesting to note that @xmath23 at the current energy range also produces track events since tau lepton can decay into muon neutrinos . in fig .
[ fig : mx_sigv_tautau ] , we summarize the sensitivities with track events for @xmath186 channel with different open angles , @xmath187 ( red solid ) , @xmath188 ( green dashed - dot ) , @xmath137 ( black dot ) , @xmath183 ( grey solid ) , and @xmath189 ( blue dashed ) .
we note that the sensitivity curves for @xmath187 and @xmath188 cross at @xmath190 gev . in other words , for @xmath191 gev , the numerator in eq .
( [ eq : sensitivity ] ) increases faster than the denominator as @xmath136 increases from @xmath118 to @xmath188 .
( see the left panel of fig .
[ fig : nevents ] ) .
moreover , we note that the sensitivity to @xmath15 can change by more than one order of magnitude as the open angle @xmath136 increases from @xmath118 to @xmath183 .
finally , with @xmath187 , icecube 5 year data can probe the pamela and ams02 positron favoured region for @xmath192 tev .
with track events for several different open angles . , scaledwidth=55.0% ]
significance on the ( @xmath4 , @xmath15 ) plane .
the @xmath6 input is taken from xenon1 t sensitivity curve .
we take @xmath150 .
the threshold energy is taken to be @xmath178 gev .
the red solid line and blue dashed line are icecube sensitivities to @xmath139 cascade events and @xmath193 track events , respectively .
, title="fig:",scaledwidth=55.0% ] + finally we discuss a pessimistic scenario that dm is not detected by xenon1 t ( a future ton - size dm detector ) . in fig .
[ fig : mx_sigv_x1 t ] , instead of using the current xenon100 limit as the input for @xmath6 , we compute the icecube 5-year @xmath28 sensitivity upper limit with the projected @xmath6 limit from xenon1 t @xcite as the input .
we set the threshold energy at 100 gev and present our result for @xmath4 up to 10 tev .
even if xenon1 t @xmath6 sensitivity limits at larger @xmath4 are weaker , only those @xmath15 arising from monochromatic annihilation channels can be probed by icecube , i.e. , by observing cascade events from @xmath139 and @xmath194 channels ( not shown on the figure ) , and by observing track events from @xmath193 channel .
however this upper limit is not stringent since the strongest bound for @xmath15 in this case is roughly @xmath195 @xmath196s@xmath197 .
in this paper , we study the neutrino signature arising from dm annihilations inside the earth core . applying icecube / deepcore effective areas , we have computed icecube / deepcore 5-year sensitivities in @xmath28 significance with track and cascade events . from the slope of sensitivity curves in fig .
[ fig : sigsip_bf ] , the neutrino event rate is more sensitive to @xmath6 than @xmath15 . to illustrate the impact of @xmath6 on the neutrino event rate , we have focused on three different scenarios according to different input @xmath6 .
hence our results can be divided by three categories .
+ we have compared the 5-year full icecube / deepcore sensitivity to @xmath6 derived from the search for dm annihilations in the earth core with the recent limit on @xmath6 by icecube 79-string detector search for dm annihilations in the sun .
we found that the small dm mass region , @xmath124 gev , can be better probed by detecting dm annihilations in the earth core . by fixing @xmath161 , we found that our @xmath186 track event result can probe the entire dama allowed region and most of the cresst - ii allowed region . if one takes the large @xmath6 favored by dama and cresst - ii as input , a rather low @xmath15 is sufficient for icecube / deepcore to achieve @xmath28 detection significance in 5 years on dm annihilations in the earth core
. it will be quite challenge for other indirect detection experiments to achieve such a sensitivity to @xmath15 in the near future .
+ we have also considered the scenario of taking the current xenon100 bound as our input @xmath6 .
we have discussed the implications by taking @xmath88 as @xmath198 gev and @xmath178 gev , respectively . in the former case
, we study the icecube / deepcore sensitivities for @xmath4 up to @xmath199 tev and consider only track events . in the latter case ,
the icecube / deepcore sensitivities are studied for @xmath4 up to @xmath198 tev with both track and cascade events . for @xmath200
gev , we found that the strongest limit of @xmath15 comes from the iron resonance region @xmath176 gev . among all the dm annihilation channels ,
the most stringent limit with track events arises from @xmath193 annihilation . for @xmath201 gev where both track and cascade events are considered , the most stringent limit of @xmath15
is from @xmath139 cascade events .
moreover , we have compared the icecube / deepcore sensitivities on the @xmath202 plane with the parameter range favored by pamela and ams02 data .
we found that both track and cascade events in @xmath23 annihilation channel can test the pamela and ams02 favored parameter space . with @xmath187 ,
the search for dm induced neutrino track events from the earth core can rule out the pamela and ams02 favoured parameter region at @xmath192 tev .
+ we finally discussed the pessimistic scenario that dm is not discovered by the future xenon1 t . with an input @xmath6 given by xenon1 t sensitivity
, we again discuss the implication on neutrino search . with @xmath101 gev and @xmath203
, we found that only those @xmath15 arising from monochromatic annihilation channels , @xmath204 , can be probed by icecube in 5 years of data taking .
however the expected bound on @xmath15 by icecube is disfavored by the current ams-02 positron flux result .
f.f.l . is supported by the grant from research and development office , national chiao - tung university , g.l.l .
is supported by national science council of taiwan under grant no .
99 - 2112-m-009 - 005-my3 and national center for theoretical sciences ( ncts ) , taiwan , and y.s.t .
is funded in part by the welcome programme of the foundation for polish science . y.s.t . also likes to thank ncts for hospitality during his visit .
r. bernabei _ et al . _
[ dama and libra collaborations ] , eur . phys .
j. c * 67 * , 39 ( 2010 ) .
c. e. aalseth _ et al . _ [ cogent collaboration ] , phys . rev .
lett . * 106 * , 131301 ( 2011 ) [ arxiv:1002.4703 [ astro-ph.co ] ] .
m. bravin _ et al . _
[ cresst- collaboration ] , astropart .
* 12 * , 107 ( 1999 ) [ hep - ex/9904005 ] .
e. aprile _ et al .
_ [ xenon100 collaboration ] , phys . rev .
lett . * 109 * , 181301 ( 2012 ) [ arxiv:1207.5988 [ astro-ph.co ] ] .
r. agnese _ et al . _
[ cdms collaboration ] , [ arxiv:1304.4279 [ hep - ex ] ] .
o. adriani _ et al . _
[ pamela collaboration ] , phys .
* 106 * , 201101 ( 2011 ) [ arxiv:1103.2880 [ astro-ph.he ] ] .
m. ackermann _ et al .
_ [ fermi lat collaboration ] , phys . rev .
d * 82 * , 092004 ( 2010 ) [ arxiv:1008.3999 [ astro-ph.he ] ] .
m. aguilar _ et al .
_ [ ams collaboration ] , phys .
lett . * 110 * , no .
14 , 141102 ( 2013 ) .
m. cirelli , m. kadastik , m. raidal and a. strumia , nucl .
b * 813 * , 1 ( 2009 ) [ arxiv:0809.2409 [ hep - ph ] ] .
j. kopp , arxiv:1304.1184 [ hep - ph ] .
a. de simone , a. riotto and w. xue , arxiv:1304.1336 [ hep - ph ] .
q. yuan , x. -j .
bi , g. -m .
chen , y. -q .
guo , s. -j .
lin and x. zhang , arxiv:1304.1482 [ astro-ph.he ] .
i. cholis and d. hooper , arxiv:1304.1840 [ astro-ph.he ] .
h. -b .
jin , y. -l .
wu and y. -f .
zhou , arxiv:1304.1997 [ hep - ph ] .
q. yuan and x. -j .
bi , arxiv:1304.2687 [ astro-ph.he ] .
m. ackermann _ et al . _
( fermi - lat collaboration ) , phys .
d * 86 * , 022002 ( 2012 ) .
d. spolyar , m. r. buckley , k. freese , d. hooper and h. murayama , arxiv:0905.4764 [ astro-ph.co ] . m. r. buckley , d. spolyar , k. freese , d. hooper and h. murayama , phys . rev .
d * 81 * , 016006 ( 2010 ) . s. k. mandal , m. r. buckley , k. freese , d. spolyar and h. murayama , phys .
d * 81 * , 043508 ( 2010 ) .
l. covi , m. grefe , a. ibarra and d. tran , jcap * 1004 * , 017 ( 2010 ) .
a. e. erkoca , m. h. reno , and i. sarcevic , phys .
d * 82 * , 113006 ( 2010 ) .
r. abbasi _ et al . _
[ icecube collaboration ] , phys .
d * 84 * , 022004 ( 2011 ) [ arxiv:1101.3349 [ astro-ph.he ] ] .
f. -f .
lee and g. -l .
lin , phys .
d * 85 * , 023529 ( 2012 ) [ arxiv:1105.5719 [ hep - ph ] ] .
f. -f .
lee , g. -l .
lin and y. -l .
s. tsai , phys .
d * 87 * , 025003 ( 2013 ) [ arxiv:1209.6226 [ hep - ph ] ] .
k. freese , phys .
b * 167 * , 295 ( 1986 ) .
a. gould , astrophys .
j. * 321 * , 571 ( 1987 ) . j. lundberg and j. edsjo , phys .
d * 69 * , 123505 ( 2004 ) [ astro - ph/0401113 ] .
l. m. krauss , m. srednicki and f. wilczek , phys .
d * 33 * , 2079 ( 1986 ) .
i. f. m. albuquerque , l. j. beraldo e silva and c. perez de los heros , phys .
d * 85 * , 123539 ( 2012 ) .
c. delaunay , p. j. fox and g. perez , jhep * 0905 * , 099 ( 2009 ) [ arxiv:0812.3331 [ hep - ph ] ] .
e. aprile [ xenon1 t collaboration ] , arxiv:1206.6288 [ astro-ph.im ] .
m. blennow , j. edsjo and t. ohlsson , jcap * 0801 * , 021 ( 2008 ) [ arxiv:0709.3898 [ hep - ph ] ] . g. l. fogli , e. lisi , a. marrone , d. montanino , a. palazzo and a. m. rotunno , phys .
d * 86 * , 013012 ( 2012 ) [ arxiv:1205.5254 [ hep - ph ] ] .
m. nauenberg , phys .
d * 36 * , 1080 ( 1987 ) .
k. griest and d. seckel , nucl .
b * 283 * , 681 ( 1987 ) [ erratum - ibid .
b * 296 * , 1034 ( 1988 ) ] .
v. berezinsky , a. bottino , j. r. ellis , n. fornengo , g. mignola and s. scopel , astropart .
phys . * 5 * , 333 ( 1996 ) [ hep - ph/9603342 ] .
g. jungman , m. kamionkowski and k. griest , phys .
* 267 * , 195 ( 1996 ) [ hep - ph/9506380 ] .
m. cirelli , n. fornengo , t. montaruli , i. a. sokalski , a. strumia and f. vissani , nucl .
b * 727 * , 99 ( 2005 ) [ erratum - ibid .
b * 790 * , 338 ( 2008 ) ] [ hep - ph/0506298 ] .
p. gondolo , j. edsjo , p. ullio , l. bergstrom , m. schelke and e. a. baltz , jcap * 0407 * , 008 ( 2004 ) [ astro - ph/0406204 ] .
t. bruch , a. h. g. peter , j. read , l. baudis and g. lake , phys .
b * 674 * , 250 ( 2009 ) [ arxiv:0902.4001 [ astro-ph.he ] ] .
r. auer , nucl .
. methods phys .
res . , sect .
a * 602 * , 84 ( 2009 ) .
m. g. aartsen _ et al . _
[ icecube collaboration ] , phys .
lett . * 110 * , 131302 ( 2013 ) [ arxiv:1212.4097 [ astro-ph.he ] ] . | we study the sensitivity of icecube / deepcore detector to dark matter annihilations in the earth core .
we focus on annihilation modes @xmath0 , and @xmath1 .
both track and cascade events are considered in our analysis . by fixing the dark matter annihilation cross section @xmath2 at some nominal values , we study the sensitivity of icecube / deepcore detector to dark matter spin - independent cross section @xmath3 for @xmath4 ranging from few tens of gev to 10 tev . this sensitivity is compared with the existing icecube 79-string constraint on the same cross section , which was obtained by searching for dark matter annihilations in the sun .
we compare this sensitivity to dark matter direct detection results as well , in particular the xenon100 ( 2012 ) limit and the parameter regions preferred by dama and cresst - ii experiments .
we also present icecube / deepcore sensitivity to @xmath5 as a function of @xmath4 by fixing @xmath3 at xenon100 ( 2012 ) and xenon1 t limits , respectively .
this sensitivity is compared with the preferred dark matter parameter range derived from the combined fitting to pamela and ams02 positron fraction data .
we conclude that the search for dark matter annihilations in the earth core provides competitive constraints on @xmath3 and @xmath5 in the case of low - mass dark matter . particularly , the expected constraint on @xmath6 for 5 years of data taking in icecube / deepcore is more stringent than the current icecube 79-string limit mentioned above .
= = = = = = = = = ==================== = ======= |
polypeptide helices are a key secondary structure motif in a wide range of proteins @xcite .
it is well known that some amino acids ( e.g. , alanine ) exhibit a stronger helix propensity than others @xcite , but the fact that the helical structure is so abundant in proteins is still intriguing . from a thermodynamic point of view , there are at least two possible limits in which helices compete with other structure prototypes . towards high temperature ,
one expects the transition to a random coil @xcite , which should become entropically favored as the temperature increases . towards low temperature , however , helices may themselves compete with other enthalpically stable conformations .
however , in the most interesting regime , namely intermediate , physiological temperatures , stability may be determined by a delicate balance between enthalpy and entropy @xcite .
we here unravel this balance quantitatively for the emergence of helical structure in a particularly well studied series of unsolvated polyalanine based peptides ac - ala@xmath0-lysh@xmath1 , @xmath2=4 - 8 @xcite .
we consider explicitly not just the helical part of conformational space , but actually the much larger , general low - energy conformational space of the peptides , of which helices are a part . in this paper
we show : ( i ) a comprehensive search of the conformational space for ac - ala@xmath0-lysh@xmath1 , @xmath2=4 - 8 ; ( ii ) harmonic free energy calculations for several structural candidates ; ( iii ) outlook on the role of anharmonicities in the potential energy surface ; and ( iv ) a theoretical comparison for longer model peptides , considering only helical motifs , with a different termination , in order to clarify the impact of lys on the soft vibrational modes .
our key finding is that there is a significant vibrational entropic stabilization of helices compared to other , more compact conformers , a contribution that should indeed make a difference in actual proteins as well .
this contribution is intrinsic to the helix and should therefore act largely independently , not entangled with environment - dependent terms such as a solvent entropy .
interestingly , an essential role of low - frequency modes is also actively debated in other areas of protein science , related to their function .
beginning with the terms that shape the potential - energy surface ( pes ) , known reasons for helix stability include @xcite ( i ) their efficient hydrogen - bond ( h - bond ) network and increasing h - bond cooperativity with helix length @xcite , ( ii ) suitably bonded and/or electrostatically favorable termination , for instance the lysh@xmath1 termination considered here @xcite , and ( iii ) remarkably , a rather specific favorable contribution of van der waals ( vdw ) interactions for @xmath3-helices @xcite .
clearly , the peptide chain length plays a role : too short chains have too few and too weak hydrogen bonds for helices to compensate the cost of strain in the backbone . in practice
, environment effects will necessarily influence helix stability @xcite . in an aqueous medium ,
the hydrophobic effect will be prominent @xcite . in water - poor conditions , like membranes , helices are frequently observed @xcite . finally , in vacuum conditions , the longer members ( @xmath48 ) of the polyalanine based peptide series studied here assume helical structure in experiment @xcite .
these helices are stable _ in vacuo _ even up to extreme temperatures ( not expected in solution ) @xcite , or after soft landing on a surface @xcite .
the potential energy surface also shapes entropy , and thus the effect of temperature @xmath5 . with increasing @xmath5 ,
the conformational entropy of the backbone will favor an unfolded state @xcite ( so - called `` random coil '' ) , while at low @xmath5 helices may also compete with other , enthalpically more stable conformers .
for instance , gas - phase ion mobility spectrometry ( ims ) by jarrold and coworkers @xcite showed that the ac - ala@xmath6-gly@xmath7-ala@xmath6h@xmath1 polypeptide is helical at @xmath5=400 k but globular at room temperature .
a similar structural change was observed in experiments involving multiply protonated polyalanine in the gas - phase in ref .
@xcite . empirical force - field based simulations by ma and coworkers @xcite of more than 60 small peptides indicate that the _ vibrational _ entropy ( harmonic approximation ) could stabilize @xmath3-helices or @xmath8-hairpins over competing low - temperature conformers .
very recently , plowright and coworkers @xcite used density - functional theory ( dft ) including dispersion contributions ( the b97-d @xcite exchange - correlation functional ) to suggest that , for a small neutral four - residue peptide , @xmath8-sheets and conformers containing 3@xmath9 helical loops are stabilized by the harmonic vibrational entropy at finite temperatures . in the present work ,
we provide independent , unambiguous , and quantitative computational evidence that the vibrational entropy acts to stabilize helical conformers with increasing temperature over more compact , enthalpically competitive structures .
the reason , in short , is traced to the softer low - frequency modes of helices , which are also reflected in the dynamics ( anharmonic case ) .
we focus on polyalanine - based peptides , since alanine is known to have a high helix propensity both in solution @xcite and _ in vacuo _ @xcite . for ac - ala@xmath0-lysh@xmath1 ( @xmath2=4 - 20 ) in the gas phase , ims @xcite and first - principles calculations
compared to experimental vibrational spectroscopy at room temperature @xcite suggest a cross - over from non - helical to helical preferred conformers as a function of polyalanine chain length .
for @xmath2=5 there is a competition between different conformers , while the @xmath2=10 and @xmath2=15 conformers are found to be firmly in the helical range @xcite . in a previous publication
@xcite we have quantified , from first principles , the contributions from electrostatics , h - bond cooperativity , and van der waals interactions on the stability of unsolvated polylalanine - based helices against unfolding .
this class of systems is thus an ideal testing ground to clarify the structural competition of non - helical ( compact ) and helical conformers as a function of chain length also toward the opposite temperature limit ( low temperature , folded state ) . in the following , we address conformational preference of ac - ala@xmath0-lysh@xmath1 _ in vacuo
_ for @xmath2=4 - 8 , i.e. , the length range in which the helical preference at room temperature develops .
our work is based on an exhaustive prediction of low - energy conformers using dft and the pbe @xcite exchange - correlation functional , corrected to account for long - range vdw interactions @xcite ( here called pbe+vdw ) .
this level of theory treats accurately , and without system - specific empirical parameters , critical length - dependent contributions such as h - bond cooperativity @xcite and vdw interactions @xcite , including their effect on vibrational frequencies and in _ ab initio _ molecular dynamics .
our conformational search strategy , used to find structure candidates of ac - ala@xmath0lysh@xmath1 , @xmath2=4 - 8 , consists of two steps . for a detailed discussion of the search strategy we use here ,
we refer the reader to ref .
@xcite . both the lys residue and the c - terminal cooh
are considered protonated throughout , as is known to be the gas - phase preference when the n - terminus is capped.@xcite in the first step , we begin by an extensive , unconstrained force field ( opls - aa @xcite ) basin - hopping search ( using the tinker package @xcite ) , aiming to simply `` list '' as many different conformers as possible . the same strategy was employed in ref .
@xcite , enumerating a huge number of candidate structures : at least 10@xmath10 conformers for each of the molecules in question .
our particular choice of the opls - aa force field is not motivated by any other reason than that an input structure `` generator '' for dft was needed .
in particular , care was taken that small changes to the parameters in the opls - aa part of the search do not affect the structure manifold considered in the dft step below.@xcite in the second step , a wide range of conformers suggested by the force field is fully relaxed using dft with the pbe+vdw exchange - correlation functional in all - electron total energy calculations ( fhi - aims program package @xcite ) .
we employ essentially converged numeric atom - centered basis sets and other numerical settings @xcite for the relaxations . in detail , the pbe+vdw part of our searches covers 1068 , 1000 , 800 , 800 , and 820 conformers for @xmath2=4 , 5 , 6 , 7 , 8 , respectively . for @xmath2=6 , 7 , and 8 , these numbers are chosen to ensure that at least all conformers identified in the lowest 7 kcal / mol ( @xmath11 0.3 ev ) of the force field step are included in the pbe+vdw relaxations . for @xmath2=4 and 5 , we explored the limits of our search strategy @xcite .
when comparing locally relaxed minima in pbe+vdw that were obtained by starting from an opls - aa local pes minimum structure , we observe a correlation of the relative energy hierarchies , but with a large scatter ( max . around 50 mev per residue ) .
therefore , a large number of conformers must be considered for post - relaxation in pbe+vdw .
the comparison also reveals structure - specific force field errors that might otherwise go unnoticed .
for example , we observe that opls - aa systematically overestimates the energy of 3@xmath9-helical structures _ in vacuo_. to exclude any unwanted impact of the overestimation of 3@xmath9-helices , we performed additional constrained basin - hopping searches in which we forced certain h - bonds of the molecule to remain 3@xmath9-helical , again followed by individual , unconstrained pbe+vdw relaxations .
the pbe+vdw relaxed conformers were sorted into `` families '' according to their h - bond pattern .
we define an h - bond to be present when an o acceptor atom is closer than 2.5 to a h donor atom . within each h - bond family
thus defined , small conformational variations are still possible , e.g. , by slight bends of the backbone atoms or different rotamers of the lysh@xmath1 side chain . typically , the lowest - energy pbe+vdw conformer is found among the family members arising within 5 kcal / mol ( 0.2 ev ) of the lowest - energy force field conformer .
harmonic vibrational frequencies and intensities were computed from finite differences .
the accuracy of the vibrational frequencies calculated in this manner is estimated by analyzing the rotational and translational modes .
these modes should be at zero frequency , and thus the deviation observed gives a limit for the accuracy of the rest of the frequencies .
we have not observed deviations of more than 2 @xmath12 in any calculations .
harmonic free energies were calculated in the harmonic oscillator / rigid body approximation .
vibrational density of states beyond the harmonic approximation were calculated from the fourier transform of the velocity time autocorrelation function , taken from _
ab initio _ molecular dynamics trajectories .
for all molecules containing @xmath2>8 alanine residues and for the li@xmath1 terminated model structures discussed in this work , no extensive conformational search was performed .
these peptides are structure models used specifically for a computer experiment to determine the development of low - frequency vibrational modes with increasing helix length for two different terminations .
their geometries are fully relaxed pbe+vdw structures .
for additional details about these calculations we refer the reader to the supporting information .
[ fig : fig1 ] summarizes the energetic ordering of the lowest - energy ( pbe+vdw ) h - bond families for @xmath2=48 , found employing our search strategy .
only the energy of the lowest energy structure belonging to each family is reported , and families are included up to 0.12 ev ( @xmath11 3 kcal / mol ) of the lowest identified minimum of the pbe+vdw pes .
@xmath3-helical conformers are highlighted in red .
the 3@xmath9-helical conformer for @xmath2=4 is highlighted in blue .
we define purely @xmath3- ( or purely 3@xmath9- ) helical conformers as those where , counting from the n - terminus , all the backbone co groups at residues @xmath13 are either connected to nh groups at residues @xmath14 ( or @xmath15 ) or to the lysh@xmath1 side chain ( usually the final three or four co groups at the c terminus ) .
coordinates and a more detailed analysis of all the geometries shown in fig .
[ fig : fig1 ] are given in the supporting information .
the conformer associated with the lowest - energy pes minimum for @xmath2=4 is rather small , connecting almost all its backbone co groups to the lysh@xmath1 termination .
the remaining h - bond at the n - terminus is bifurcated , comprising an @xmath3- and a 3@xmath9-helical h - bond .
this conformer could therefore be classified as the smallest possible @xmath3-helical prototype in this series .
in contrast , the lowest - energy pes minima for @xmath2=5 and 6 are _ not _ simple helices .
each contains an `` inverted '' h - bond where one co group points to the n - terminus and its connecting nh group points to the c - terminus , producing more compact structures . in fig .
[ fig : fig1 ] , these bonds are highlighted in yellow and pointed to by an arrow .
for @xmath2=5 , we have previously denoted this conformer as `` g-1 '' @xcite . for @xmath2=7 and 8
, the lowest - energy pes minima correspond to @xmath3-helices . in each case , they are closely followed by a conformer that we characterize as compact / globular ( families 2 of @xmath2=7 and 8 , with an energy separation of 20 and 60 mev respectively ) .
thus , we already observe a cross - over with peptide length to @xmath3-helical lowest - energy minima of the pes at @xmath2=7 .
however , based on the energy hierarchies of the structures of the local pes minima alone , one would not expect a purely helical ensemble of conformers at room temperature at @xmath2=7 or 8 .
simple boltzmann factors would indicate a mix of structure candidates . yet
, the experimental work by kohtani and jarrold @xcite does indicate a complete room - temperature structural cross - over at @xmath2=8 at the latest , albeit based on a completely different line of reasoning ( water adsorption behavior of `` helical '' versus `` globular '' conformers ) . for the low - energy conformers in fig .
[ fig : fig1 ] , we also compute and show in the same figure the impact of the vibrational free energy at room temperature ( @xmath5=300 k ) in the harmonic approximation .
remarkably , the relative stability of the @xmath3-helices is systematically enhanced by the vibrational free energy contribution for _ all _ @xmath2 . for @xmath2=5 and 6
the @xmath3-helical conformers move down in energy with respect to the ( non - helical ) lowest energy pes minimum . for @xmath2=7 and 8
the @xmath3-helices now become the isolated minima . in detail , we observe : + - for @xmath2=8 , the energetic interval between the @xmath3-helical lowest energy conformer and the nearest globular one ( red shaded areas ) now amounts to 0.14 ev .
the additional family 1a at 0.13 ev is another @xmath3-helix with a slightly modified terminating h - bond network . with the vibrational free energy included , the energy hierarchy is thus consistent with the experimental claim that @xmath3-helices dominate over all other possible conformers in gas - phase experiments for @xmath2=8 @xcite .
+ - for @xmath2=7 , the same qualitative picture emerges . here , the next remaining conformer ( family 3 ) at 300 k is a mixed 3@xmath9/@xmath3-helix . the competing compact conformers
( representatives of family 2 and 2a ) are significantly destabilized by @xmath16(@xmath5 ) .
+ - for @xmath2=6 , the @xmath3-helix emerges as the room - temperature minimum free energy conformer , but the competing non - helical pes minimum , family 1 , remains close ( 50 mev ) .
+ - for @xmath2=5 , the difference between the @xmath3-helix and family 1 ( g-1 ) decreases by 60 mev at 300 k , compared to the pes minimum , but family 1 remains overall more favorable .
+ in essence , there is an uniform stabilization of helical over more compact ( globular ) structures as the temperature increases .
the stabilization tendency increases with peptide length , confirming quantitatively and systematically from first principles the related observations in refs .
we next demonstrate that it is indeed the _ vibrational entropy _
term which is critical , and then pinpoint the physical reason among the low - frequency vibrational modes . in fig .
[ fig : fig3 ] we analyze the individual quantities composing the vibrational free energy differences between : families 1 and 2 of @xmath2=4 , families 1 and the @xmath3-helical conformers of @xmath2=5 and 6 , as well as families 1 ( @xmath3-helical ) and 2 for @xmath2=7 and 8 .
the energy terms plotted are the pbe+vdw pes minimum energy , the harmonic internal energy ( containing the zero - point energy ) @xmath17 , and the entropy term @xmath18 .
they are reported in fig .
[ fig : fig3 ] as energy differences , taking the non - helical conformer of each @xmath2 as the reference , such that negative slopes correspond to a stabilization of the @xmath3-helices . upon inspection of fig . [
fig : fig3 ] , we observe a monotonic stabilization of all helical conformers with increasing @xmath5 .
while for the shortest molecule ( @xmath2=4 ) there is hardly any observable effect , the stabilization trend is enhanced with increasing length . for @xmath2=6
, we predict a cross - over of the lowest - energy structures at @xmath19 150 k. it is clear that among the individual contributions to the vibrational part of the free energy @xmath20 , the entropy term @xmath21 is indeed always the most important helix - favoring term .
the zero - point - energy ( @xmath22 at @xmath5=0 ) , is also favorable , but on a smaller scale .
= 4 ; families 1 ( g-1 ) and 2 ( @xmath3 ) of @xmath2=5 ; families 1 ( g-1 " like ) and 3 ( @xmath3 ) of @xmath2=6 ; families 1 ( @xmath3 ) and 2 ( compact ) of @xmath2=7 ; and families 1 ( @xmath3 ) and 2 ( compact ) of @xmath2=8 . in each case
, the non-@xmath3-helical conformer was taken as the reference ( i.e. negative slopes mean @xmath3-helix stabilization ) .
the pbe+vdw energy differences are plotted with a yellow dash - dash - dot line , the harmonic internal energy @xmath23 as black dash - dot lines , the entropy term @xmath24 as red dashed lines , and the sum @xmath16 as green full lines .
[ fig : fig3],scaledwidth=60.0% ] the observed stabilization of the @xmath3-helical conformers can be understood by analyzing the lowest - energy vibrational modes of different conformations . for simple helices ,
these modes have been estimated and analyzed before @xcite .
the important point here is that we have available a direct comparison against many other , non - helical structure candidates .
these modes dominate in the harmonic free energy expression at room @xmath5 , so that conformers with lower - frequency modes will effectively be stabilized over others . in table
[ tab : all - first - modes ] , the frequencies ( in @xmath12 ) corresponding to the first normal modes of all conformers shown in fig .
[ fig : fig1 ] are reported .
the lowest - energy @xmath3-helices , marked in red in table [ tab : all - first - modes ] , show first vibrational normal modes between 8 @xmath12 and 13 @xmath12 .
the same is true for families 1a of @xmath2=7 and 8 , which only differ in details of the termination .
in contrast , the conformers that have first vibrational modes of at least 20 @xmath12 are all compact non - helical , for example : 28 @xmath12 for family 2 of @xmath2=7 , and 22 @xmath12 for family 1 ( g-1 motif ) of @xmath2=5 . the g-1 motif has the first vibrational mode lying around 20 @xmath12 for @xmath2=5 , 6 , and 7 ( marked with a * * * symbol in table [ tab : all - first - modes ] ) . of course
, the energy destabilization of the compact conformers with respect to the helices will depend not only on the first vibrational mode but also on the overall distribution of low - frequency modes . to illustrate this point
, we show the frequencies of vibration lying between 0 and 50 @xmath12 for families 1 ( @xmath3 ) and 2 ( globular ) of @xmath2=7 and 8 in fig .
[ fig : fig4](a ) and ( b ) .
the @xmath3-helices have lower first vibrational normal modes , _ and _ have slightly higher ( one or two modes ) densities of modes in this region than the globular conformers .
.position of the lowest vibrational mode , in @xmath12 , for the lowest energy conformers of each family shown in fig .
[ fig : fig1 ] for @xmath2=4 - 8 .
we use red characters to indicate the @xmath3-helical conformers , a * * * symbol for the g-1-like conformers of @xmath2=5 and 6 , and blue characters for the 3@xmath9-helical conformer .
the numerical accuracy of all computed frequencies is 2 @xmath12 or better ( see supplementary material ) .
[ tab : all - first - modes ] [ cols="^,^,^,^,^,^",options="header " , ] here , a comment regarding structures that are more elongated than @xmath3-helices is in order .
it is customary to compare the stability of @xmath3-helices with 3@xmath9 helices and @xmath8-sheets or fully extended structures ( e.g. , in references @xcite and many others ) .
following our rationale above , the more extended the structure is , the softer the low vibrational modes will be .
this indeed happens for the most extreme case , the fully extended structure ( fes ) , where we find the first vibrational mode to lie around only 2 @xmath12 ( calculated for @xmath2=8 and 15 ) .
in fact , entropically stabilized @xmath8-sheets in _ neutral _ polyalanine in the gas - phase have been suggested in ref .
for 3@xmath9 helices , in our own structure searches for @xmath2=4 - 8 , we find the first vibrational mode to lie always very close to their @xmath3-helical counterpart .
accordingly , it is the enthalpic energy difference that favors @xmath3-helices specifically over 3@xmath9 for all molecules studied . for infinite periodic structures , calculation of phonons and vibrational free energies for @xmath3- , 3@xmath9- , and @xmath25-helices , and the fes in ref .
@xcite corroborate our results .
there , all helices are destabilized with respect to the fes at 300 k. the @xmath25-helix , which is the most compact among the helices studied in ref .
@xcite , is most destabilized by the vibrational entropy term . for ( a ) family 1 ( @xmath3 ) and family 2 ( globular ) of ac - ala@xmath7-lysh@xmath1 and
( b ) family 1 ( @xmath3 ) and family 2 ( globular ) of ac - ala@xmath26-lysh@xmath1 .
( c ) vibrational density of states taken from the velocity autocorrelation function of 21ps of microcanonical aimd at @xmath27 300 k , for family 1 ( @xmath3 ) and family 2 ( globular ) of ac - ala@xmath26-lysh@xmath1 .
light colored bars in ( c ) correspond to the harmonic normal frequencies of vibration [ same as ( b ) ] .
[ fig : fig4],scaledwidth=50.0% ] we next show that the observations discussed in the last paragraph should hold also in a fully anharmonic picture . even at relatively low temperature , where the structure still stays generally constant ,
we expect first the inherent anharmonicity of the local , nearly harmonic pes , then the lowest - barrier transitions between basins ( side chain rotations ) , and then transitions between locally different backbone conformations and h - bond networks @xcite to contribute .
unfortunately , a direct calculation of these terms ( e.g. , by thermodynamic integration ) is computationally prohibitive in dft .
we can , however , use explicit _ ab initio _ molecular dynamics simulations to gain some insights . in fig .
[ fig : fig4](c ) , we compare the fourier - transformed velocity autocorrelation functions of the family 1 ( helical ) and family 2 ( compact non - helical ) conformers of @xmath2=8 , extracted from explicit microcanonical _ ab initio _ molecular dynamics simulations ( 21 ps total time , 1 fs time step , initially thermalized to approximately room temperature ) .
the fourier transform of the velocity autocorrelation function corresponds to a vibrational density of states ( vdos ) . at @xmath5=0 and for classical nuclei
, the autocorrelation function should reflect only the harmonic eigenmodes of fig .
[ fig : fig4](b ) . compared to these modes ,
the onset of the calculated vdos at @xmath5@xmath11300 k is noticeably shifted towards lower frequencies for both conformers , but the onset frequency for the helix is still significantly lower than for the non - helical structure .
thus , the lower vibrational frequencies of the helix are also carried over to the full ( anharmonic ) motion of the conformers .
in addition , the integral over the vdos up to 50 @xmath12 is 4% larger for the @xmath3-helical family 1 than for the compact family 2 , i.e. , the general downshift of frequencies in this region is also preserved .
regarding the overall local structure stability during the aimd simulation , we observe that the h - bond pattern of family 2 ( compact non - helical ) stays essentially the same throughout the entire simulation ( see figure s.1 in the supporting information ) . in contrast
, family 1 displays local structural fluctuations in the helical part , occasionally forming short - lived 3@xmath9-like h - bond connections .
similar fluctuations also occur in simulations of longer helices ( e.g. , ac - ala@xmath28-lysh@xmath1 in ref .
it seems plausible that the overall greater `` floppiness '' of the helix compared to the more compact non - helical h - bond network adds another favorable entropic contribution at room temperature .
this direct observation that we make has been guessed as the motive for the loss in entropy observed in @xmath3-helical formation , compared to a ppii helical structure in ref .
@xcite . according to our reasoning
, the ppii structure , being less compact than the @xmath3-helices should indeed show more fluctuations , just like @xmath3-helices do if compared to more compact conformers .
finally , we show that , especially for the short peptides considered here , there are two independent aspects to the stabilization of helical conformers : length and termination . to isolate their role ,
we examine the character of the lowest - frequency modes as a function of peptide length ( also for longer helices ) for two different terminations .
the first is the lysh@xmath1-terminated series , which is the main subject of this work .
we contrast this series with li@xmath1-terminated polyalanine molecules ala@xmath0-li@xmath1 , a much more rigid termination as we shall see . for the latter ,
we assume @xmath3-helices for all @xmath2 ( fully relaxed in dft - pbe+vdw ) , which are at least locally stable when li@xmath1 is placed in contact with the last three residues at the c terminus . in fig .
[ fig : termination_deform](a ) , we compare the position of the first and second vibrational modes of these @xmath3-helical geometries of ala@xmath0-li@xmath1 , @xmath2=5 - 15 , and 20 , with the @xmath3-helical geometries of ac - ala@xmath0-lysh@xmath1 , @xmath2=5 - 15 , and 19 . for both peptide series ,
there is a monotonic decrease of the first and second vibrational frequencies for @xmath297 , but the starting point is much higher for the li@xmath30 termination ( 26 @xmath12 at @xmath2=7 ) than for the lysh@xmath30 termination . a softening trend of the respective modes in neutral polyalanine helices with increasing length has also been observed in ref . @xcite . in order to characterize this vibrational mode in more detail , fig .
[ fig : termination_deform](b ) visualizes the displacement of the backbone atoms of ac - ala@xmath31-lysh@xmath1 when deforming this molecule along the first vibrational mode . fig .
[ fig : termination_deform](c ) shows the relative length changes of the hydrogen bonds in the structure upon deformation along this mode .
subfigures ( d ) and ( e ) show the equivalent data for ala@xmath32-li@xmath1 .
for both molecules , the vibration spans the helical part of the structures . for the lysh@xmath1 terminated molecule ,
we show that the actual lysh@xmath30 termination ( connected to the last four co residues ) is clearly involved in the vibration .
in contrast , the hypothetical li@xmath1 charged termination constrains especially the c terminus to be much more rigid , as evidenced by the almost zero change of all li - o distances . here
, the n - terminus is much more involved in the lowest - frequency modes .
the same trends are observed for the smaller molecules in both series .
movies containing 3d visualizations of the first vibrational modes for helical and compact structures are contained in the supporting information .
we thus conclude two points : + ( i ) helices are entropically favored by allowing delocalized , soft low - frequency modes that we do not observe in competing , more compact conformers of the same lysh@xmath1 termination ( evidenced by table [ tab : all - first - modes ] ) .
+ ( ii ) for short conformers , the already electrostatically favorable lysh@xmath1 termination @xcite is additionally helpful by allowing soft , delocalized modes to include the termination also for short conformers , in contrast to the hypothetical , much harder charged termination by li@xmath1 . for long enough helices
these softer low frequency modes should exist regardless of the termination .
in summary , we show from first principles , quantitatively , and for a particularly well studied series of polyalanine peptides how helices emerge with length and temperature as the leading structural pattern from a vast array of possible competing conformers .
the crossover to helical stability with length is already apparent based on local structural minima of the potential - energy surface alone , due to the critical role of h - bond networks including their cooperativity @xcite as well as that of vdw terms @xcite .
in addition , the contribution from softer low - frequency vibrational modes acts to stabilize helices with increasing temperature over their more compact competition .
the specific experimental claim @xcite of exclusively helical conformers at and above @xmath2=8 is thus explained by _ both _ enthalpic and entropic effects acting together at finite temperature . _
ab initio _ molecular dynamics simulations corresponding to approximately room temperature suggest that these trends are further strengthened by anharmonic effects .
the emergence of room - temperature helix stability with length in ac - ala@xmath0-lysh@xmath1 is thus the result of a subtle balance of enthalpic and entropic terms . in a non - vacuum environment , further terms
would obviously contribute , but we expect the fact that helices , in general , allow locally softer vibrational modes to hold . regarding the overall ubiquity of the helical motif in folded peptides and proteins , at least , we here show that low - frequency modes will be a significant quantitative contribution .
the authors would like to acknowledge dr .
carsten baldauf for numerous helpful discussions and suggestions about the figures and paper in general , including the schematic backbone vibration visualization in fig .
[ fig : termination_deform ] .
additional computational details , xyz geometries of conformers discussed in this paper , detailed information about the structures discussed in this paper , and movies illustrating the first vibrational modes of helices and compact conformers .
rossi , m. _ ab initio study of alanine - based polypeptide secondary - structure motifs in the gas phase _ , ph.d .
thesis ( fritz - haber - institut der max - planck - gesellschaft and technische universitt berlin ) , * 2011 * , http://opus.kobv.de / tuberlin / volltexte/2012/3429/. | helices are a key folding motif in protein structure .
the question which factors determine helix stability for a given polypeptide or protein is an ongoing challenge .
here we use van der waals corrected density - functional theory to address a part of this question in a bottom - up approach . we show how intrinsic helical structure is stabilized with length and temperature for a series of experimentally well studied unsolvated alanine based polypeptides , ac - ala@xmath0-lysh@xmath1 . by exploring extensively the conformational space of these molecules
, we find that helices emerge as the preferred structure in the length range @xmath2=4 - 8 not just due to enthalpic factors ( hydrogen bonds and their cooperativity , van der waals dispersion interactions , electrostatics ) , but importantly also by a vibrational entropic stabilization over competing conformers at room temperature .
the stabilization is shown to be due to softer low - frequency vibrational modes in helical conformers than in more compact ones .
this observation is corroborated by including anharmonic effects explicitly through _ ab initio _ molecular dynamics , and generalized by testing different terminations and considering larger helical peptide models . |
in the preceding paper@xcite we suggested a new method for ab - initio electronic - structure calculations of solids .
the main steps of this procedure are : 1 .
perform a conventional , self - consistent , band - structure calculation for an effective one - particle hamiltonian , namely , the hartree hamiltonian , to obtain a suitable basis set of bloch functions . 2 . by taking into account only a finite number @xmath0 of bands
one chooses a truncated one - particle hilbert space .
the marzari - vanderbilt@xcite algorithm is then used to construct a maximally localized set of wannier functions , which span the same truncated one - particle hilbert space .
3 . all one - particle ( tight - binding ) and two - particle ( coulomb ) matrix elements of the hamiltonian within this wannier function basis are calculated .
the resulting electronic many - body hamiltonian in second quantization with parameters determined from first principles is studied within standard many - body approximations for lattice electron systems . in this paper
we apply this scheme to the 3d transition metals fe , co , ni , and cu .
we use the `` linear muffin - tin orbital '' ( lmto ) method within the `` atomic - sphere approximation '' ( asa)@xcite to perform the band - structure calculation for the hartree hamiltonian in first quantization .
the direct coulomb matrix elements of the maximally localized wannier basis are rather large , about 20 ev in magnitude .
we then use the simplest possible many body approximation , the hartree - fock approximation ( hfa ) , to study the second - quantized multiple - band hamiltonian .
our results are compared with those obtained from a standard lsda calculation.@xcite although the 3d - bands and the 4s - band overlap in the l(s)da approximation , our unscreened hfa calculations give 3d - bands that lie considerably lower ( between 10 and 20 ev ) than the 4s - band .
the hfa correctly predicts ferromagnetism for the ferromagnetic metals fe , co , and ni and no magnetism for cu , but with a much larger exchange splitting between majority and minority 3d bands than obtained within lsda and with a slightly larger magnetic moment per site than obtained experimentally or within lsda . on the other hand ,
the total energy is lower in hfa than in lsda .
the lsda results for metals are probably more reliable than our new hfa results , which lack important screening and correlation effects . in order for our method to go beyond lsda
we would need to use better many body methods than the ( unscreened ) hfa , which should be possible within our scheme .
our purpose in presenting the hfa results is to demonstrate that the new ab - initio many - body method that we have proposed is feasible and can be applied to practical calculations of materials .
in addition , because hfa is the assumed standard starting point before adding complex many - body correlations , and because it is the simplest many - body approximation ( or the best mean - field one - particle approximation ) , it is useful to know what the hfa predicts for the 3d transition metals , which have been so heavily studied by other techniques .
comparisons of hfa results with experiments , other ( higher order ) theories , or established standard methods such as the l(s)da should demonstrate the effects of correlation on electronic properties in d - electron systems . to the best of our knowledge we do not know of any published hfa results ( band structure , density of states , magnetism , magnetic moment , total energy , etc . ) for the 3d ferromagnets fe , co and ni , unless it was implicitly applied to these materials for schemes like the local ansatz@xcite , where hfa results serve as an input to higher order calculations .
the is not surprising since the hfa has , from very early on , been viewed as a poor approximation for metals . for example , when applied to the homogeneous electron gas ( as the simplest model of an infinite metallic system ) , the hfa has well - known fermi edge singularities@xcite .
these lead , in particular , to a vanishing density of states ( dos ) at the fermi energy , which is , of course , unphysical .
this unphysical feature usually prevails in actual hfa - calculations for real metals@xcite , though sometimes this singularity is hard to see in actual hfa - results@xcite . in our calculations
the non - locality is handled through the calculation of expectation values ( matrix elements of the density matrix ) , which makes hf calculations as easy as hartree calculations .
furthermore , because of our localized wannier basis , we only keep on - site and a few inter - site coulomb and exchange matrix elements .
hence our calculations have an effective short - ranged coulomb interaction . although longer - range coulomb matrix elements are small in our calculations , which is why we truncate them , it is possible that if all of them were kept to infinite distances that they could add up to give fermi edge singularities ( which are due to the long - ranged nature of the bare coulomb interaction ) and other standard anomalies .
correlation or screening would quickly kill these effects .
the paper is organized as follows . in section [ sec : lmtohartree ] we describe the lmto - hartree calculations and the hartree - results for the band structure and density of states , respectively .
section [ sec : wannierfunctions ] describes some results obtained for the maximally localized wannier functions obtained within the marzari - vanderbilt algorithm@xcite , in particular their localization properties .
results for the matrix elements , in particular the direct coulomb and exchange matrix elements are given in section [ sec : matrixel ] ; we also compare these results with calculations of the slater integrals .
the application of the ( unscreened ) hfa to the multiband many - body hamiltonian in second quantization is the subject of section [ sec : hartreefock ] . for an interpretation of the results we compare the numerical hfa results obtained for the crystal with previous atomic hfa results and with numerical and analytical results for a simplified local ( atomic or zero band width ) model in section [ sec : atomichfa ] .
a comparison with the more standard lsda - results follows in section [ sec : lsda ] , before the paper closes with a short discussion .
for the four materials of interest ( fe , co , ni , cu ) we performed a selfconsistent hartree band - structure calculation . besides the nuclear charge we used the ( experimentally ) known results for the lattice structure ( bcc for fe , fcc otherwise ; co should actually be hexagonal ) and for the lattice constant as input . for the band - structure calculation we used the lmto method@xcite within the atomic sphere approximation ( asa ) .
the radius of the ( overlapping ) muffin - tin spheres ( the wigner - seitz radius @xmath1 ) is determined by the condition that the sphere volume equals the volume of the unit cell .
this yields the following values for the wigner - seitz radius : @xmath2 for fe , @xmath3 for co , @xmath4 for ni and @xmath5 for cu ( ref . ) . within the muffin - tin spheres
the potential and wave functions are expanded in spherical harmonics with a cutoff @xmath6 , i.e. , s , p , and d - orbitals are included . within the lmto -
asa the eigenfunctions , i.e. the bloch wave functions ( inside a muffin - tin sphere ) , are given in terms of the solution to the radial schrdinger equation @xmath7 to some fixed energies @xmath8 and its energy derivative @xmath9 : @xmath10 where the @xmath11 denote the ( complex ) spherical harmonics and @xmath12 is the band index .
total density of states ( of both spin directions ) around the fermi level ( chosen as zero of the energy axis @xmath13 ) obtained from the selfconsistent hartree calculation for fe . ] the same as in fig .
[ fig : feha ] for co. ] the same as in fig .
[ fig : feha ] for ni . ] the same as in fig .
[ fig : feha ] for cu . ] figs .
[ fig : feha ] - [ fig : cuha ] show the nonmagnetic ( spin - degenerate ) hartree results for the density of states ( dos ) .
one observes that the five narrow 3d - bands are similar in energy and therefore hybridize with the 4sp - bands .
for fe , co , and ni the fermi level @xmath14 falls in the midst of the partially filled 3d - bands , and hence there is a large dos at the fermi level . for cu @xmath14
is above the filled 3d bands , where the states have more 4sp - like character and the dos is small ( free electron like ) .
the bloch functions obtained from this hartree calculation are used as the basis of a truncated hilbert space . in the band calculation
, there is already a natural truncation due to the spherical harmonics expansion cutoff @xmath15 , which limits the calculation to 9 bands for one atom per unit cell .
the connection between the bloch functions in the different muffin - tin spheres is given by the standard bloch relation @xmath16 hence the bloch function in a single muffin - tin sphere determines the function for the whole crystal .
however , this simple relation holds for bloch functions only . any other function ( to be represented as linear combination of bloch functions )
can , nevertheless , be decomposed into its contributions within the different muffin - tins according to @xmath17 where @xmath18 for @xmath19 .
we now turn to the construction and description of the wannier functions and will present results for their main properties for the four materials under investigation . because @xmath20 it is sufficient to investigate the wannier functions only for one arbitrarily chosen lattice vector @xmath21 : @xmath22 .
this wannier function has contributions not only in the muffin - tin sphere around @xmath21 but also in other muffin - tin spheres , and we can use the decomposition ( [ eq : p3 ] ) into the contributions from the different muffin - tin spheres @xmath23 after application of the marzari - vanderbilt algorithm , which is briefly summarized in the preceding paper@xcite and in more detail in ref .
, the new set of bands that are used to calculate the maximally localized wannier functions can no longer be classified by pure angular momentum quantum numbers .
the wannier functions are rather admixtures having different @xmath24-contributions ( 3d , 4s , 4p etc . ) .
but , since the original bloch functions from which the wannier functions are constructed were given in terms of a spherical harmonics expansion , the new wannier functions ( and their contribution in each individual muffin - tin sphere ) can also be decomposed into these spherical harmonics contributions @xmath25 one can then calculate the weight of the contributions to the wannier function ( centered at @xmath21 ) within the different muffin - tin spheres @xmath26 and one can also decompose this into the different @xmath27-contributions according to : @xmath28 for the 3d - system iron these quantities are tabulated in table [ tab : fe_wf ] .
the first line is the weight @xmath29 in the center muffin - tin . between 88 and 98% of the total weight of the wannier functions
is to be found already within the center muffin - tin ; this shows how well localized our wannier functions are with the lowest five functions having values of more than 95% .
rows 24 in this table indicate the different @xmath27-contribution or @xmath27-character of the wannier functions .
one sees that the optimally localized wannier functions are not pure within their @xmath27-character , but the lowest five wannier functions ( 0 - 4 ) still have mainly @xmath30 ( 3d ) character .
higher band - index states ( which are slightly less well localized according to row 1 ) are admixtures that have mainly @xmath31 ( 4p ) character ( about 50 % ) , but also a considerable amount of @xmath32 ( 4s ) and @xmath30 ( 3d ) character .
corresponding results for the other 3d - systems co , ni , and cu are similar and , therefore , not repeated here .
.[tab : fe_wf ] some properties of the lowest eight maximally localized wannier functions of fe .
[ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] we have also evaluated the slater integrals@xcite : @xmath33 where @xmath34 is a radial ( atomic ) d - wave function ( obtained by solving the schrdinger equation for a radial symmetric potential , for instance ) . note that only the three integrals @xmath35 , @xmath36 and @xmath37 are required to determine all the coulomb @xmath38-matrix elements . using the radial d - wave function obtained from the hartree calculation
we obtain the following values for the slater integrals of the four 3d - systems : @xmath39 21.62 ev for fe , 23.18 ev for co , 24.69 ev for ni , and 26.27 ev for cu .
this means , the slater integrals @xmath35 are rather good estimates of our ( averaged ) coulomb matrix elements .
these values are also in agreement with older results obtained in calculations for 3d - atoms@xcite . in table
[ tab : f^k ] we show our @xmath40-values for the four 3d - crystals and compare them with corresponding atomic calculations from ref . .
obviously , there is fairly good agreement between these atomic and our results .
after we have determined the matrix elements within our restricted basis set of 9 maximally localized wannier functions ( per site and spin ) , we have a hamiltonian in second quantization of the form @xmath41 for which all the matrix elements are known from first principles .
the simplest approximation one can now apply is the hfa , which replaces the many - body hamiltonian by the effective one - particle hamiltonian @xmath42 { \langle}c_{3\sigma'}^{\dagger } c_{4\sigma ' } { \rangle}~~.\end{aligned}\ ] ] here the expectation values @xmath43 have to be determined selfconsistently for the hf hamiltonian ( [ eq : hfhamilton ] ) .
note that the fock ( exchange ) term is spin ( @xmath44 ) dependent and may , therefore , give rise to magnetic solutions .
( unscreened ) hartree - fock band - structure of fe ; the full line shows the majority ( spin up ) , the dashed line the minority spin component . ] density of states ( per spin direction ) for fe within the ( unscreened ) hartree - fock approximation ; the full line shows the majority ( spin up ) , the dashed line the minority spin contribution . ] the same as fig .
[ fig : fehfband ] for co. ] the same as fig .
[ fig : fehfdos ] for co. ] the same as fig .
[ fig : fehfband ] for ni . ] the same as fig .
[ fig : fehfdos ] for ni . ] the same as fig .
[ fig : fehfband ] for cu . ] the same as fig .
[ fig : fehfdos ] for cu . ]
the hartree - fock results for the four materials of interest are shown in figs .
[ fig : fehfband][fig : cuhfdos ] .
we show the effective hf band structure and its density of states ( dos ) . in our hfa calculations
there are no singularities ( or a vanishing dos ) at the fermi level since we start from a localized description and consider the coulomb matrix elements only locally ( on - site and for a few neighbor shells ) .
therefore , we implicitely truncate the coulomb interaction in real space and in practice work with an effective short - ranged interaction . within hfa
the main part of the 3d - bands lies between 18 and 22 ev below the fermi level and is separated from the 4sp - bands .
we find magnetism in hfa for fe , co , and ni in agreement with experiment .
the five majority spin d - bands are about 20 ev below the fermi energy and are completely filled .
but the partially filled minority d - bands have two ( for fe ) , three ( for co ) , and four ( for ni ) filled bands between -18 and -15 ev , and the rest are around and above the fermi level .
this results in magnetic moments per atom ( in units of the bohr magneton @xmath45/atom ) of 2.9 for iron , 1.9 for cobalt , and 0.76 for nickel . for copper
no magnetism and exchange splitting of the 3d - bands is obtained , but the ( spin degenerate ) 3d - bands are at about 22 ev below the fermi level and separated from the 4sp - bands . if we compare these results with the results of the simple hartree approximation shown in figs .
[ fig : feha][fig : cuha ] we see that the exchange term has two effects : it produces an exchange splitting and the possibility of magnetic solutions , and it draws the 3d - bands energetically down by an amount of about 20 ev .
compared with experiment the hfa overestimates magnetism and leads to overly large values for the magnetic moment per site ( about 1030% too large ; experiment gives magnetic moments of about 2.2 for iron , 1.7 for co , and 0.6 for ni ) .
this is consistent with heisenberg or ising model studies where the mean - field approximation hfa also has the tendency to overestimate magnetism and magnetic solutions .
non - magnetic hfa - band structure for co. ] total density of states ( of both degenerate spin directions ) for co obtained within the non - magnetic hfa - solution . ]
however , the reason why the 3d - bands lie so far below the fermi level and the 4sp - band in hfa has nothing to do with the existence and overestimation of magnetism .
this can be seen already from the non - magnetic system cu , for which the ( fully occupied ) 3d - bands also lie at about 22 ev below the fermi level ( see figs . [
fig : cuhfband ] and [ fig : cuhfdos ] ) . to demonstrate this also for a system with a partially filled 3d - band
we have done a non - magnetic hartree - fock - calculation for co ( by forcing equal occupation for both spin directions ) .
the results for the band structure and the dos are shown in figs .
[ fig : conm - band],[fig : conm - dos ] .
we observe again that the main part of the 3d - bands are well below the 4s - bands and fermi level ; note the hybridization gap caused by the unoccupied 3d - bands above the fermi level .
we have seen in the previous section that one effect of the hfa calculation , when compared with the hartree calculation , is the shift of the 3d - bands down ( about 20 ev below the fermi level and about 810 ev below the bottom of the 4sp - band ) .
this shift of the d - bands is about the same energy as the coulomb matrix elements @xmath46 , and roughly agrees earlier atomic hartree - fock calculations@xcite , where the 3d - states are also about 10 ev below the 4s - states .
energy eigenvalues from quasi - atomic hfa - calculation .
the numbers in brackets indicate the degeneracy . ] because the inter - site hopping matrix elements in table [ tab : ujtaverage ] are much smaller than the u - values one may consider an expansion in @xmath47 , with the zeroth order approximation to completely neglect hopping . doing this
, we have performed a quasi - atomic hfa calculation for co , by including only the on - site one - particle and two - particle ( coulomb ) matrix elements .
the results are summarized in fig .
[ fig : coatom ] ; the degeneracy of the different levels is also indicated . in the paramagnetic case
, we find that the 3d - bands are below the 4s - bands ( at the fermi level ) by about 6 to 7 ev , which is in rough agreement with the earlier atomic hfa results@xcite .
the splitting between the occupied and unoccupied 3d - states is about 23 ev , which is the on - site @xmath46 for co. magnetic hfa solutions are also found in the atomic limit for co , as shown in the right panel of fig .
[ fig : coatom ] .
the majority - spin 3d states ( @xmath48 and @xmath49 ) are now completely filled and energetically lie lower than the corresponding non - magnetic hfa - states .
but only the ( 3-fold degenerate ) @xmath48-states of the minority - spin electrons are filled whereas the @xmath49-states of the minority electrons are empty ( and now even 26 ev above the occupied d - states ) .
the additional energetical shifts between the occupied 3d - states in the paramagnetic and ferromagnetic atomic hfa solution are due to the exchange matrix elements @xmath0 .
this behavior can qualitatively be understood within the framework of the following simple , analytically solvable model .
similar to the numerical hfa - results presented and discused above , we neglect all intersite one - particle ( hopping ) and interaction matrix elements . furthermore , we assume that we have diagonalized the one - particle hamiltonian , taking into account only the atomic 3d - levels and assuming that the on - site one - particle diagonal matrix elements @xmath50 , the coulomb matrix elements @xmath46 , and the exchange matrix elements @xmath0 are equal , i.e. , that the 3d - levels are degenerate in the atomic limit with no crystal - field effects .
then the atomic part of the many - body hamiltonian can be written as @xmath51 where @xmath52 denote the 5 ( degenerate ) 3d - states .
the standard hartree - fock decoupling leads to @xmath53 \\ & - & j \sum_{j
\neq i } \langle c_{j\sigma}^{\dagger } c_{j\sigma } \rangle \big ) c_{i\sigma}^{\dagger } c_{i\sigma } ~~.\end{aligned}\ ] ] here we have assumed that the hartree - fock hamiltonian has the same symmetry as the uncorrelated hamiltonian , and hence off - diagonal expectation values @xmath54 for @xmath55 vanish . from this equation it is clear that the hf hamiltonian can be written in terms of an effective one - particle energy @xmath56 where @xmath57 - j \sum_{j \neq i } \langle c_{j\sigma}^{\dagger } c_{j\sigma } \rangle .\ ] ] in the simple hartree approximation ( ha ) the exchange decouplings are neglected , which means that all the decoupling terms with the negative sign would not occur . therefore , the corresponding hartree one - particle energies are given by @xmath58 comparing this result with the hartree - fock one - particle energies , we find that the hf occupied levels are shifted downwards by an amount of @xmath59 relative to the hartree levels . momentarily setting @xmath60 , we see that for @xmath61 occupied levels the hartree approximation gives the one - particle energies @xmath62 whereas the hfa yields @xmath63 the occupied hartree - fock one - particle energies are lower than the corresponding hartree one - particle energies by @xmath46 , which is a consequence of the artificial and unphysical self - interaction still present in the hartree approximation that is exactly cancelled in hartree - fock .
this also explains why the hartree - fock bands are shifted downwards from the hartree bands by an energy of the amount @xmath46 .
one also sees from this simple atomic - limit hartree - fock model that the energy difference between the highest occupied and the lowest unoccupied effective hartree - fock one - particle energies is again essentially @xmath46 , which is once more in agreement with our numerical results for the crystal and for the atom ( cf .
fig.[fig : coatom ] ) .
note that we have ignored @xmath64 interactions , which cause an additional shift of d - bands below the s - bands by about an additional 10 ev in the full hfa calculations .
taking into account the exchange interaction @xmath0 again and denoting by @xmath65 the number of occupied states with spin @xmath44 ( i.e. @xmath66 ) one obtains in hfa @xmath67 then the total energy in hfa is given by @xmath68 for the total energy we have added the necessary correction term to the sum of the occupied energy levels ( much like the double counting term that shows up in band - structure calculations ) .
now for partially filled 3d - shells the occupation of the different spin directions may be different . denoting @xmath69 we obtain for the total energy @xmath70 the magnetic ( @xmath71 ) total
energy is lower than the nonmagnetic ( consistent with hund s rules ) .
take once more co with 8 3d - electrons .
the paramagnetic ( nonmagnetic ) state has the occupations @xmath72 ( @xmath61=8 and @xmath73=0 ) . for this configuration ( corresponding to the left panel in fig .
[ fig : coatom ] ) one obtains @xmath74 the hund s rule magnetic solution has 3d - states of one spin - direction completely filled , i.e. , @xmath75 and @xmath76 ( @xmath61=8 and @xmath73=2 ) .
this gives @xmath77 therefore , the magnetic configuation ( with a magnetic moment of 2 for the atom ) is energetically more favorable by @xmath0 .
note also the exchange splitting in the occupied energy eigenvalues @xmath78 and that our model would predict the unoccupied minority spin @xmath79 state to be @xmath80 higher in energy than the corresponding occupied majority spin state .
the simple model in this section differs from the results shown in fig .
[ fig : coatom ] in that we have replaced the full matrix of @xmath46 and @xmath0 by scalar values for d - states only ( ignoring s - d interactions , for example )
. however , it captures all of the important physics without attempting to be completely quantitative .
for comparison with the hfa results described in section [ sec : hartreefock ] we have also performed a standard local spin - density approximation ( lsda ) , ab - initio , band - structure calculation with the lmto - asa method .
the results for the dos are shown in figs .
[ fig : felsda][fig : culsda ] . for the magnetic systems fe , co , and
ni one obtains also an exchange splitting and the prediction of magnetic solutions with magnetic moments of 2.18 for iron , 1.58 for co , and 0.58 for ni , which are in better agreement with experiment than the hfa results .
this is consistent with the expectation that l(s)da is tuned to accurately predict ground - state properties .
density of states ( per spin direction ) for fe within the ( unscreened ) local spin - density approximation(lsda ) ; the full line shows the majority ( spin up ) , the dashed line the minority spin contribution . ]
the same as fig .
[ fig : felsda ] for co. ] the same as fig .
[ fig : felsda ] for ni . ]
total density of states ( of both degenerate spin directions ) for cu within the local density approximation(lsda ) ] the energy spectra of the bands ( dos ) are quite different from the hfa .
for example , the 3d - bands now fall into the same energy region as the 4sp - bands , i.e. , the lsda - results are not so different from the hartree - results .
this means that the exchange - correlation energy leads only to a small shift of the 3d - bands downwards by at most a few ev and a smaller exchange splitting ( also of the magnitude of 1 ev ) in the case of magnetic solutions . on the other hand , in the lsda calculations the self - interaction terms
are not completely canceled , i.e. , an ( unrealistic ) self - interaction is included , which may lead to 3d bands that lie energetically too high , as discussed for the atomic limit in the previous section .
total density of states for co within the exchange - only local density approximation(lsda ) . ] to see the effect of correlations within lsda , we have also performed an exchange - only calculation for co. the result is shown in fig .
[ fig : coxolsda ] . obviously the ( majority ) d - bands lie lower than the ones in the full lsda - calculation shown in fig . [
fig : colsda ] .
but the shift is of the magnitude of 1 ev only , i.e. , very minimal when compared with the large drop in the full hfa . on the other hand , this exchange - only lsda - result also does not contain self - interaction corrections , which are responsible for the large shift downwards of the d - bands .
nevertheless , the lsda result indicates that a possible effect of correlations is to shift the 3d - bands up relative to exchange - only calculations , and hence one would expect a similar effect if correlations could be added to the full hfa - calculations . total ground state energy ( of the valence electrons ) obtained in hartree - approximation , l(s)da and hfa for the 3d transition metals fe through cu .
] one can also calculate the total energy in the hartree , hfa , and lsda approximations .
the results obtained for the four materials of interest are shown in fig .
[ fig : etot ] .
we see that the total energy is always significantly lower in hfa than in the hartree approximation , which is expected because the hfa minimizes the total energy .
the hfa total energy is also lower than the l(s)da , and the lsda result is lower than the simple hartree result .
because of the unknown approximations that go into constructing l(s)da , it is hard to guess ahead of time that this would be the case .
however , it is well known that the l(s)da approximation produces a bad total exchange - correlation energy ; the reason why such good agreement with experiment is found is that relative exchange - correlation energies are nonetheless reasonably accurately calculated .
in this paper we have presented the results of ( unscreened ) hfa calculations for the 3d transition metals fe , co , ni and cu . we obtain magnetic solutions for fe , co , and ni with ( slightly ) too large magnetic moments when compared to experimental or lsda results .
the occupied hfa 3d - bands lie about 20 ev below the fermi level ( and the hartree result ) , which is also the magnitude of the splitting between occupied and unoccupied 3d - bands and of the magnitude of the on - site coulomb matrix element ( the `` hubbard '' u ) .
this downwards shift of the hfa 3d - bands compared to the hartree- and l(s)da-3d - bands can be understood as due to the self - interaction correction of hfa .
one may argue that these results are not surprising and an artifact of using the unscreened hfa .
our ab - initio calculation of the direct coulomb matrix elements yields large values of the magnitude of 20 ev .
therefore the coulomb energies are large for these materials and hfa can be considered to be an approximation for the selfenergy which is correct only in linear order in the coulomb interaction .
but for these large values of the u - terms hfa is certainly not sufficient but one has to apply better many - body approximations
. one should apply systematic extensions of hfa , which within the standard perturbational approach can be represented by ( a resummation of an infinite series of ) feynman diagrams , or one can try to apply the recently so successful non - perturbational many - body schemes like `` dynamical mean field theory '' ( dmft)@xcite or variational ( gutzwiller ) approaches@xcite .
the simplest standard diagram series are the bubble diagrams leading essentially to the random phase approximation
this means just a renormalization of the interaction line , i.e. the pure `` naked '' coulomb interaction has to be replaced by a `` dressed '' interaction . or in other words , the exchange ( fock ) contribution has not to be calculated with the bare coulomb matrix elements but with screened coulomb matrix elements .
probably the non - perturbational schemes like dmft are also only applicable for screened coulomb matrix elements . to summarize , the new many - body ab - initio approach proposed in ref .
is applicable to real materials .
maximally localized wannier functions and their one- and two - particle matrix elements can be calculated from first principles so as to obtain a many - body multi - band hamiltonian in second quantization .
standard methods of many - body theory can then be applied to this hamiltonian .
the simplest approximation ( hfa ) is unreliable , because the large coulomb matrix elements are unscreened .
therefore , in the future better many - body approximations should be used . within the standard feynman diagram approach the most straightforward next step consists in a summation of bubble diagrams leading to a renormalized ( screened ) coulomb interaction .
this requires calculating the exchange contribution not with the bare but with a screened coulomb interaction . to take into account the effects of screening one
has to calculate the charge susceptibility and the ( static ) dielectric constant , which can be done within a generalized lindhard theory , for instance .
this work has been supported by a grant from the deutsche forschungsgemeinschaft no .
cz/31 - 12 - 1 .
it was also partially supported by the department of energy under contract w-7405-eng-36 .
this research used resources of the national energy research scientific computing center , which is supported by the office of science of the u.s .
department of energy under contract no .
de - ac03 - 76sf00098 .
i. schnell , g. czycholl , r.c.albers , phys .
b , preceeding paper ( 2003 ) n. marzari and d. vanderbilt , phys . rev .
b * 56 * , 12847 ( 1997 ) .
skriver , _ the lmto method _ ( springer - verlag , heidelberg 1984 ) r.m .
dreizler , e.k.u .
gross , _ density functional theory _ , springer ( berlin , heidelberg , new york 1990 ) h. eschrig , _ the fundamentals of density functional theory _
( teubner stuttgart , leipzig 1996 ) g. stollhoff , phys .
b * 58 * , 9826 ( 1998 ) g.d .
mahan , _ many - particle physics _
( plenum press new york 1990 ) c. pisani , r. dovesi , c. roetti , _ hartree - fock ab initio treatment of crystalline systems _ ( lecture notes in chemistry , vol .
48 , springer - verlag berlin heidelberg 1988 ) ; c.pisani , _ quantum - mechanical ab - initio calculation of the properties of crystalline materials _ , lecture notes in chemistry , vol .
67 , spinger verlag , heidelberg , 1996 h.j .
monkhorst , phys .
b * 20 * , 1504 ( 1979 ) r.dovesi , c. pisani , f.ricca , c.roetti , phys . rev .
b * 25 * , 3731 ( 1982 ) o.k .
andersen , phys .
b * 12 * , 3060 ( 1975 ) i. schnell , g. czycholl , r.c .
albers , phys .
b * 65 * , 075103 ( 2002 ) j.c .
slater , phys . rev . * 34 * , 1293 ( 1929 ) r.e .
watson , phys .
rev . * 118 * , 1036 ( 1959 ) ; * 119 * , 1934 ( 1959 ) l. hodges , r.e.watson , h.ehrenreich , phys . rev .
b * 5 * , 3953 ( 1972 ) for a review see : a. georges , g. kotliar , w. krauth , m.j .
rozenberg , rev .
phys . * 68 * , 13 ( 1996 ) w. weber , j. bnemann , f. gebhard , in _ band - ferromagnetsim _
( lecture notes in physics , vol .
580 , p.9 ( eds . : k. baberschke , m. donath , w. nolting , springer berlin 2001 ) ; j. bnemann , w. weber , f. gebhard , phys .
b * 57 * , 6896 ( 1998 ) | the ab - initio many - body method suggested in the preceding paper is applied to the 3d transition metals fe , co , ni , and cu .
we use a linearized muffin - tin orbital calculation to determine bloch functions for the hartree one - particle hamiltonian , and from these obtain maximally localized wannier functions . within this wannier basis
all relevant one - particle and two - particle coulomb matrix elements are calculated . the resulting second - quantized many - body hamiltonian with ab - initio parameters
is studied within the simplest many - body approximation , namely the unscreened , selfconsistent , hartree - fock approximation ( hfa ) .
we present these hfa results , which we believe are the first to have been done for crystalline 3d transition metals , and compare them with those obtained from the standard local ( spin ) density approximation ( lsda ) within density functional theory ( dft ) .
although the d - bands sit considerably lower within hfa than within l(s)da , the exchange splitting and magnetic moments for ferromagnetic fe , co , and ni are only slightly larger in hfa than what is obtained experimentally or within lsda .
the hfa total energies are lower than the corresponding l(s)da calculations . |
the nature of unidentified high - energy @xmath0-ray sources in the galaxy was one of the major unanswered questions at the end of the egret era .
the third egret catalog contained 170 unidentified sources , 74 of which were at galactic latitude @xmath10@xcite .
rotation - powered pulsars are believed to dominate the galactic @xmath0-ray source population ( e.g. @xcite ) , but their visibility is linked to their beam patterns . soon after launch , the _ fermi gamma - ray space telescope _ began to unveil many 3eg sources , discovering the radio - quiet pulsar in the cta1 supernova remnant associated with 3eg j0010 + 7309 @xcite , detecting the radio - loud pulsar psr j2021 + 3651 ( associated with 3eg j2021 + 3716 ) @xcite , seen independently by _ agile _
@xcite , as well as the radio pulsar psr j1028@xmath25819 , associated with 3eg j1027@xmath25817 @xcite and new populations of radio - quiet @xmath0-ray pulsars , detectable using blind search techniques @xcite . for the pulsars detected in @xmath0 rays ,
the bulk of the electromagnetic power output is in high energies .
the @xmath0-ray emission is thus crucial for understanding the emission mechanism which converts the rotational energy of the neutron star into electromagnetic radiation
. the discovery of many new @xmath0-ray pulsars will provide strong constraints on the location of the @xmath0-ray emitting regions , whether above the polar caps @xcite , or far from the neutron star in the so - called outer gaps " @xcite , or in the intermediate regions like the slot gap " @xcite having two - pole caustic " geometry @xcite .
in this paper we report the _ fermi _ detection of the two pulsars psr j1048@xmath25832 and psr j2229 + 6114 , which have spin characteristics similar to other young pulsars typified by the vela pulsar .
@xcite provide a discussion of `` vela - like '' pulsars .
+ psr j1048@xmath25832 ( b1046@xmath258 ) is located in the carina region at low galactic latitude ( @xmath11 , @xmath12 ) .
it was discovered during a 1.4 ghz parkes survey of the galactic plane and has a period @xmath13123.7 ms @xcite .
it has a spin - down luminosity @xmath14 of @xmath15ergs@xmath8 , for a moment of inertia @xmath16 of @xmath17g@xmath18 , a surface dipole magnetic field strength of @xmath19 g , and a characteristic age @xmath20 of 20kyr .
high - resolution observations of the coincident @xmath21 source by the _ chandra x - ray observatory _ and _ xmm - newton _ revealed an asymmetric pulsar wind nebula ( pwn ) of @xmath22 , surrounding a point source coincident with the pulsar but so far no x - ray pulsations were detected @xcite .
psr j1048@xmath25832 has been proposed as a counterpart of the steady egret source 3eg j1048@xmath25840 @xcite , as suggested by positional coincidence , spectral , and energetic properties .
detailed analysis of the egret data found possible @xmath0-ray pulsations at @xmath23400 mev , a double - peaked light curve with @xmath24 peak phase separation ( fig.[fig : j1048 - 5832_lc_multi_energy ] , @xcite ) .
an hi distance determination for the pulsar yields between 2.5 and 6.6 kpc @xcite .
the ne2001 model @xcite assigns a distance of 2.7 kpc based partly on the hi distance determination . for this paper
we adopt 3 kpc as the distance to the pulsar .
+ psr j2229 + 6114 is located at ( @xmath25,@xmath26 ) = ( 106.6,2.9 ) within the error box of the egret source 3eg j2227 + 6122 @xcite . detected as a compact x - ray source by _ rosat _ and _ asca _ observations of the egret error box , it was later discovered to be a radio and x - ray pulsar with a period of @xmath27 = 51.6 ms @xcite .
the radio pulse profile shows a single sharp peak , while the x - ray light curve at 0.8 10 kev consists of two peaks , separated by @xmath28 .
_ agile _ recently reported the discovery of @xmath0-ray pulsations above 100 mev @xcite .
the pulsar is as young as the vela pulsar ( characteristic age @xmath29kyr ) , as energetic ( @xmath30ergs@xmath8 ) , and is evidently the energy source of the `` boomerang '' arc - shaped pwn g106.65 + 2.96 , suggested to be part of the supernova remnant ( snr ) g106.3 + 2.7 discovered by @xcite .
recently , the pwn has been detected at tev energies by milagro @xcite .
studies of the radial velocities of both neutral hydrogen and molecular material place the system at @xmath31 800 pc @xcite , while @xcite suggest a distance of 3 kpc estimated from its x - ray absorption .
the pulsar dm , used in conjunction with the ne2001 model , yields a distance of 7.5 kpc , significantly above all other estimates . for this paper
we again adopt a distance of 3kpc .
the large area telescope ( lat ) aboard _ fermi _ is an electron - positron pair conversion telescope sensitive to @xmath0-rays of energies 0.02 to @xmath32300 gev .
the lat is made of a high - resolution silicon microstrip tracker , a csi hodoscopic electromagnetic calorimeter , and segmented plastic scintillators to reject the background of charged particles @xcite . compared with its predecessor egret
, the lat has a larger effective area ( @xmath318000 @xmath18 on - axis for e @xmath32 1 gev ) , improved angular resolution ( @xmath33 0.5@xmath34 at 1 gev for events in the front section of the tracker ) , and a higher sensitivity ( @xmath35@xmath7s@xmath8 ) 10@xmath36 @xmath7 s@xmath8 sr@xmath8 ( @xmath320.1 gev and a photon spectral index of -2.1 with no spectral cutoff ) . ] .
the large field of view ( @xmath312.4 sr ) allows the lat to observe the full sky in survey mode every 3 hours .
the lat timing is derived from a gps clock on the spacecraft and @xmath0-rays are hardware time - stamped to an accuracy significantly better than 1 @xmath37s @xcite .
the lat software tools for pulsars have been shown to be accurate to a few @xmath37s @xcite . in this paper ,
the data used for the spectral analysis were collected during _ fermi _ s first - year all - sky survey , beginning 2008 august 4 . for the timing analysis
, we added data collected during the commissioning phase observations from 2008 june 30 to 2008 august 3 , that included pointed observations of the vela pulsar and other targets . the data for psr j1048@xmath25832 end 2009 april 10 , while those for psr j2229 + 6114 end 2009 march 23 .
only @xmath0 rays in the diffuse " class events ( the tightest background rejection ) were selected .
in addition , we excluded those coming from zenith angles @xmath32 105 , where @xmath0-rays resulting from cosmic - ray interactions in the earth s atmosphere produce an excessive background contamination .
psr j10485832 is observed in the radio band at approximately monthly intervals since early 2007 at the 64-m parkes radio telescope in australia @xcite .
a typical observation is of 2 min duration at a frequency of 1.4ghz with occasional additional observations at 0.7 and 3.1ghz .
full details of the observing and data analysis can be found in @xcite .
the pulsar is known to have glitched in the past @xcite and indeed glitched just prior to the launch of fermi @xcite .
like many high @xmath38 pulsars , psr j1048@xmath25832 is highly polarized in the radio band over a wide frequency range @xcite .
+ the timing solution uses 20 pulse times of arrival ( toas ) and was derived using the pulsar timing software tempo2 @xcite .
the timing solution fits for the pulsar s spin frequency and frequency derivative and also whitens the data using the so - called fitwaves algorithm within tempo2 .
the resulting rms of 0.287 ms is a substantial improvement on the pre - whitened solution .
note that the extra fitwaves parameters are not supported by the standard _
software tools but were included in this analysis .
measurements of the dispersion delay across the 1369mhz band with a bandwidth of 256 mhz lead to a dispersion measure ( dm ) of 128.822 @xmath39 0.008pc@xmath40 , with no indication that the dm is varying over time .
this dm is used to correct the radio toas to infinite frequency for phasing the @xmath0-ray photon toas .
psr j2229 + 6114 is being observed at the nrao green bank telescope ( gbt ) @xcite and the lovell telescope at jodrell bank @xcite .
the rotational ephemeris used here to fold @xmath0-ray photons is based on toas obtained from both telescopes between 2008 june 17 and 2009 march 23 .
there are 25 such toas from gbt with average uncertainty of 0.2ms , each from a 5 minute observation mainly at a central frequency of 2.0ghz .
the 44 jodrell bank toas have an average uncertainty of 0.3ms , obtained from individual 30 minute observations at 1.4ghz . in our timing
fits with tempo , we fixed the position at that known from x - ray observations @xcite .
psr j2229 + 6114 shows some timing noise over the 9 month interval , requiring a fit to rotation frequency @xmath41 and its first two derivatives .
in addition , on mjd @xmath42 a small glitch occurred , with fractional frequency step of @xmath43 and @xmath44 .
the post - fit rms is 0.24ms .
we have measured the dm with separate sets of gbt observations at three widely spaced frequencies , obtaining @xmath45pc@xmath40 .
this is used to correct 2ghz arrival times to infinite frequency , for comparison with the @xmath0-ray profile .
the events were analyzed using the standard software package _ _ science tools _ _ ( st ) for the _ fermi _ lat data analysis and tempo2 .
the timing parameters used in this work will be made available on the servers of the _ fermi _ science support centerfermi.gsfc.nasa.gov/ssc/data/access/lat/ephems/ ] .
for the timing analysis of psr j1048@xmath25832 , we selected @xmath0-rays with energy @xmath320.1 gev within a radius of 10 around the radio pulsar position . then , we applied an energy - dependent angular radius cut @xmath46 @xmath47degrees , keeping all the photons included in a radius of 035 .
this selection approximates the lat point spread function ( psf ) and maximizes the signal - to - noise ratio over a broad energy range .
we corrected photon arrival times to the solar system barycenter using the jpl de405 solar system ephemeris @xcite .
the event times were then folded with the radio period using the parkes ephemeris . the bin - independent _
h_-test @xcite results in a probability @xmath48 that the modulation would have occurred by chance . this value is more than 4 orders of magnitude smaller than the previous egret results @xcite , establishing firmly this source as a @xmath0-ray pulsar .
figure [ fig : j1048 - 5832_lc_multi_energy ] ( top panel ) shows the resulting 50-bin @xmath0-ray phase histogram above 0.1 gev .
the two peaks appear asymmetric , with a slow rise and a fast fall .
we fit each peak with two half - lorentzian functions , i.e. with different widths for the leading and trailing sides .
p1 ( @xmath49 ) is sharper with a full width at half maximum ( fwhm ) of @xmath50 , while p2 ( @xmath51 ) is a bit broader with an fwhm of @xmath52 .
the 1.4 ghz radio profile is shown at the bottom panel for comparison .
the @xmath0-ray light curve consists of two peaks , with p1 at phase @xmath53 and p2 at phase @xmath54 , leading to a phase separation @xmath55 of @xmath56 .
the errors are respectively the @xmath0-ray fit uncertainty and that caused by the dm uncertainty .
these results are in agreement with those found by @xcite .
we observe structure related to a shoulder or a `` bridge '' region between 0.17 and 0.30 in phase , which trails the first @xmath0-ray peak .
note that the profile is very similar to the vela light curve , which consists of both two peaks separated by 0.43 in phase and a bridge region @xcite .
we also defined the `` off - pulse region '' as the pulse minimum ( @xmath57 ) . to check this assumption , we estimated the background represented by the dashed line ( 73 counts / bin ) from a ring with 1@xmath582 surrounding the source .
nearby sources are removed , and we normalized to the same phase space as our selection .
the result is in good agreement with the off - pulse region . as a consequence
, the total number of pulse photons from the pulsar is estimated at @xmath59 , with a background contribution of @xmath60 events .
figure [ fig : j1048 - 5832_lc_multi_energy ] ( middle panels ) shows the 50-bin @xmath0-ray phase histograms of psr j1048@xmath25832 in four energy bands ( 0.1 0.3 gev , 0.3 1gev , 1 3gev , @xmath323gev ) .
evolution in the light curve shape with energy is visible , although more data are needed to constrain the peak widths .
notably , below 0.3gev the first peak seems wider than at high energies .
this feature could be explained by the contamination of a lat source ( 0fgl 1045.6@xmath25937 ) less than 1@xmath61 from the pulsar .
we also looked for an evolution of the ratio p1/p2 with energy , as seen by egret for vela , crab , geminga , psr b1951 + 32 @xcite and seen by the _ fermi _
lat for vela @xcite and psr j0205 + 6449 @xcite . in each energy band , we calculated the peak height with respect to the background level as estimated from the off pulse interval .
the ratio shows no variation with a confidence level of @xmath3169@xmath62 .
this is not a particularly stringent bound , and we expect to accumulate much more data to eventually detect a possible variation of p1/p2 with energy .
finally , note that between 1 and 3 gev , the trailing shoulder p1 ( @xmath63 = 0.17
0.33 ) is significant and there is still evidence of it for e @xmath32 3 gev , whereas above 3 gev the two peaks are still observed with the highest energy photon detected in p1 at 19gev . for the timing analysis of psr j2229 + 6114
, we used the same energy - dependent selection criteria as for psr j1048@xmath25832 .
@xmath0-ray toas were folded with the radio period using the gbt and lovell telescope ephemeris .
figure [ fig : j2229 + 6114_lc_multi_energy ] ( top panel ) shows the resulting 50-bin histogram of folded counts above 0.1 gev compared with the phase - aligned radio pulse profile ( bottom panel ) .
the profile shows a single , asymmetric peak ( @xmath64 ) , that has been fit by a two half - lorentzian function .
the fit places the peak at @xmath65 with fwhm of @xmath66 .
we estimated the background level from a 1
2@xmath61 ring around the pulsar during the 0.65 0.15 pulse minimum .
this is represented by the dashed line ( 48 counts / bin ) .
the result is consistent with the off - pulse region . as a consequence
, the total of pulsed photons from the pulsar is estimated at @xmath67 , with a background contribution of @xmath68 events . to examine the energy - dependent trend of the @xmath0-ray pulse profile ( figure [ fig : j2229 + 6114_lc_multi_energy ] , middle panels ) , phase histograms
are plotted over 4 energy intervals ( @xmath69gev , @xmath70gev , @xmath71gev and @xmath323 gev ) , showing a possible drift of the peak . between 0.1 0.3 gev ,
the peak is offset from the radio pulse by @xmath72 according to a two half - lorentzian fit , while the offsets for 0.3 1gev , 1
3gev and @xmath32 3 gev are @xmath73 , @xmath5 and @xmath74 , respectively .
the peak positions are compatible between 0.1 3 gev , and a slight mis - alignment appears above 3gev .
however , the data can not constrain strongly the pulse phase dependence , even if the mis - alignment between the x - ray and @xmath0-ray profile suggests a spectral energy dependence of the high energy light curve .
finally , we note that the highest energy photon is 10.8gev at phase @xmath31 0.30 . in figure
[ fig : j2229 + 6114_lc_multi_energy ] we also show the 110kev x - ray profile of psr j2229 + 6114 obtained from an _ xmm _ observation on 2002 june 15 ( mjd 52440 ) with effective exposure time of 20ks .
the data were folded using a contemporaneous ephemeris based on jodrell bank and gbt observations .
the highest peak of the x - ray profile lags the radio pulse by @xmath63 = @xmath75 .
analysis of an _ rxte _ observation from mjd 52250 yields a consistent radio
x - ray offset . there is no energy dependence of the x - ray pulse shape within the 110kev range , while the sharpness of the peaks as well as their spectral shape indicates that the emission is predominantly non - thermal . in order to obtain the phase - averaged spectrum for psr j1048@xmath25832 and psr j2229 + 6114 ,
a maximum likelihood spectral analysis @xcite was performed , using the lat tool ` gtlike'@xmath76 .
we used the `` pass 6 v3 '' instrument response functions ( irfs ) , which are a post - launch update to address @xmath0-ray detection inefficiencies that are correlated with trigger rate .
the systematic errors on the effective area are @xmath77 5% near 1 gev , 10% below 0.1 gev and 20% over 10 gev .
a circular region around 15@xmath61 from the source was modeled including nearby lat sources around the source position .
the galactic diffuse background was accounted for by using maps based on the galprop model called 54_59xvarh7s @xcite .
we modeled the isotropic background through a tabulated spectrum derived from a fit to lat data at high galactic latitude , freezing the galactic model and including the detected point sources within a radius of 15 .
we fit the spectrum of psr j1048@xmath25832 with a power law with an exponential cutoff between 0.1 and 0.7 in phase , that can be described by the equation : @xmath78 with @xmath79 in gev , the term @xmath80@xmath7s@xmath8gev@xmath8 , a spectral index @xmath81 = @xmath82 and a cut - off energy @xmath83gev .
the first uncertainty is statistical , while the second is systematic . from this result
, we obtained an integral photon flux in the range 0.1 - 100 gev of @xmath6@xmath7s@xmath8 , which is about one third of the flux of the egret source 3eg j1048@xmath25840 .
figure [ fig : sed_j1048 ] shows both the overall fit between 0.1 and 20 gev ( solid line ) , and the spectral points obtained for 6 logarithmically - spaced energy bins and performing spectral analysis in each interval , assuming a power law shape for the source . to check the assumption of a cut - off energy in the spectrum
, we also modeled the pulsar with a simple power law of the form @xmath84 .
the chance probability to incorrectly reject the hypothesis of a pure power law spectrum is @xmath31 10 @xmath85 .
the pulsar is listed in the bright source list of the _ fermi _ lat @xcite as 0fgl j1047.6@xmath25834 , which is located at ( ra , dec ) = ( 161.922 , @xmath258.577 ) with a 95% confidence level radius of 0.138@xmath61 .
there is another lat point source located @xmath311@xmath61 away , 0fgl j1045.6@xmath25937 .
we modeled this unidentified source with a simple power law .
the best fit result gives a spectral index @xmath86 and an integral photon flux of @xmath87@xmath7s@xmath8 .
the sum of the fluxes of this source plus psr j1048@xmath25832 is @xmath31 6.7 @xmath8810@xmath89 @xmath7s@xmath8 which is consistent with the flux of 3eg j1048@xmath25840 of @xmath90@xmath7s@xmath8 .
the egret source 3eg j1048@xmath25840 as well as the cos b source 2cg 28800 @xcite were apparently made up of these two sources , which have now been resolved by the _ fermi _ lat .
initially , a 15 degree circular region around the source position and the same galactic diffuse background used for psr j1048@xmath25832 were modeled to fit psr j2229 + 6114 .
however , some structures around the object were not taken into account in the galactic model , and overestimate the flux of the pulse regions .
we finally adopted a model for the galactic diffuse emission based on six galactocentric ring maps of n(h i ) and w(co ) and on the spatial distribution of the inverse compton intensity modeled by galprop .
this intensity as well as the gamma - ray emissivities per ring were adjusted to maximize agreement with the observations , taking into account detected point sources of gamma rays .
this approach is similar to that described by @xcite for modeling egret data , and is being used by the lat team to fit the model of galactic diffuse emission for the first public release .
+ psr j2229 + 6114 , referenced as the lat point source 0fgl j2229.0 + 6114 in the bright source list , was modeled by a power law with a simple exponential cut - off ( eq . [ eq : spectrum ] ) , between 0.15 and 0.65 in phase .
figure [ fig : sed_j2229 ] ( solid line ) shows the phase - averaged spectral energy distribution from the likelihood fit , with @xmath91 = @xmath92@xmath7s@xmath8gev@xmath8 , a spectral index @xmath81 = @xmath93 and a cut - off energy @xmath94gev . from this
, we estimated an integral photon flux of @xmath9@xmath7s@xmath8 .
this value is 10% lower than the flux of @xmath95@xmath7s@xmath8 of the egret source 3eg j2227 + 6122 obtained for a power law with a spectral index of @xmath96 and larger than the flux of @xmath97@xmath7s@xmath8 measured by agile @xcite .
we also fit the lat data to a simple power - law model , yielding a flux of @xmath98@xmath7s@xmath8 and a spectral index of @xmath99 .
note that the spectral model using an exponential cut - off is better constrained with a difference between the log likelihoods of @xmath100 , rejecting the power - law hypothesis . for both psr
j10485832 and psr j2229 + 6114 the statistics available is not enough to significantly rule out a super - exponential cutoff in favor of a simple exponential cutoff .
as a first search for unpulsed emission from the boomerang nebula , we fitted a point - source to the off - pulse data at the radio pulsar position in the energy band 0.2 100 gev .
no signal was observed from the pwn . after scaling to the full pulse phase
, we derived a 95% confidence level upper limit on the flux of @xmath101@xmath7s@xmath8 .
the @xmath0-ray light curves of psr j1048@xmath25832 and psr j2229 + 6114 cover a wide range in phase , suggesting that the @xmath0-ray beams cover a large solid angle .
this in turn seems to favor the outer magnetospheric emission models , in particular the outer gap ( og , @xcite ) and the slot gap ( sg , @xcite ) models . the @xmath0-ray luminosity is given by @xmath102 where @xmath103 is the observed energy flux at the earth line of sight ( at angle @xmath104 to the rotation axis ) , @xmath105 is the pulsar distance , and @xmath106 is the beaming correction factor that depends on the geometry of the emission pattern .
the factor @xmath106 is a function of the pulsar magnetic inclination @xmath107 , and is model sensitive .
it is given by @xcite : @xmath108 where @xmath109 is the radiated flux as a function of the viewing angle @xmath110 and the pulsar phase @xmath63 .
in the ratio @xmath111 , the numerator is the total emission over the full sky , and the denominator is the expected phase - averaged flux for the light curve seen from earth . for the polar cap model @xcite , the @xmath0-ray emission originates at a few stellar radii from the surface , implying an emission with a small solid angle , that is , @xmath112 1 . for both og and sg models , where the emission is far away from the neutron star
, the resulting @xmath111 values can be near unity , and even exceed 1 when the dominant sampling of the flux @xmath113 arises for viewing angles @xmath110 quite disparate from @xmath104 . for psr j1048@xmath25832 , the observed integral energy flux obtained by integrating equation [ eq : spectrum ] times
the energy is @xmath114erg@xmath7s@xmath8 , leading to a @xmath0-ray luminosity of @xmath115 @xmath111 ( @xmath105/3 kpc)@xmath116 and an efficiency @xmath117 = @xmath118 = 0.10 @xmath111 ( @xmath105/3 kpc)@xmath116 .
the peak separation in the @xmath0-ray light curve and the @xmath0-ray efficiency are useful to place constraints in the now - favored outer magnetospheric emission models . using the @xmath0-ray light curve `` atlas '' of @xcite
we can estimate an allowed range @xmath119 60 85 , @xmath120 70 80 , and @xmath121 0.7 1.1 for the og model , and @xmath119 50 75 , @xmath120 50 75 , and @xmath121 1.1 for the sg ( two - pole caustic ) model . the variation of the radio polarization position angle for a pulsar constrains the impact angle @xmath122 = @xmath107 - @xmath123 .
fitting the rotating vector model @xcite to the psr j1048@xmath25832 radio data , a value of @xmath122 smaller than 10 has been derived @xcite .
both theoretical models thus have a good range of possible solutions , but these estimates assume the efficiency relation @xmath124 of @xcite , which gives @xmath125 0.02 , about a factor of five smaller than that derived for this pulsar from its inferred luminosity . assuming our measured quantity @xmath117 = 0.10 the allowed range remains about the same for the two - pole caustic ( tpc ) model and the range of @xmath107 shifts to @xmath31 70 90 for the og model . the phase lag between the radio pulse and the first @xmath0-ray pulse
is in agreement with both og and tpc or sg models . + for psr j2229 + 6114 , the observed integrated energy flux is @xmath126erg@xmath7s@xmath8 .
thus , we estimate a @xmath0-ray luminosity @xmath127 = 2.6@xmath8810@xmath128(@xmath105/3 kpc)@xmath116 erg s@xmath8 , and deduce an efficiency @xmath129 = 0.011 @xmath111 ( @xmath105/3 kpc)@xmath116 , which is a factor 1.6 larger than the efficiency estimated by the relation @xmath124 .
note that with the estimated snr distance of 0.8kpc the efficiency decreases to 0.001 @xmath111 , while with the distance derived from the dm ( 7.5kpc ) the efficiency increases to 0.07 @xmath111 , emphasizing the importance of the distance determination . as for psr j1048@xmath25832 , we can compare light curves with the geometrical models of @xcite to derive constraints on the geometry . for
outer magnetosphere models with @xmath130 , both tpc and og geometries can deliver single @xmath0-ray pulses at a range of angles .
single pulse solutions appear for @xmath131 , @xmath132 for the tpc / sg geometry and over the range @xmath133 , @xmath134 and @xmath121 0.47 - 1.08 for the og model . for psr j2229
+ 6114 , we have an additional geometrical constraint from modeling of the x - ray pwn torus surrounding the pulsar , as measured by _
chandra _ ; @xcite fit these data to determine a viewing angle @xmath135 .
this is in the overlap range consistent with both the tpc and the og single pulse solutions .
finally , although there is some uncertainty in determining the precise phase of the magnetic axis from the radio pulse , the lag of the strong @xmath0-ray peak from the radio peak ( @xmath136 ) provides additional information . examining the sample light curves in figures 9 & 10 of @xcite we find that single peaks at this phase are found only when they can be identified as the normal ` p2 ' component , with a ` bridge ' of emission to earlier phases and ` p1 ' faint or missing .
such light curves appear only for small efficiency @xmath137 and only for a modest range of angles : @xmath138 , @xmath139 ( tpc / sg , with p1 faint ) or @xmath140 , @xmath141 ( og , with p1 absent ) .
this is precisely the @xmath110 range required by the x - ray torus fitting ; including the small @xmath122 constraint from the detection of radio emission , we see that the radio , x - ray and @xmath0-ray data are all consistent with @xmath142 , @xmath143 , implying @xmath144 for both models .
thus low @xmath0-ray efficiency , high altitude emission and a well constrained viewing geometry together provide a consistent picture for gamma - ray light curve shape . on the other hand , the gamma - ray light curve does not seem to match the the clearly double - peaked x - ray light curve .
however , we have argued that the gamma - ray light curve is intrinsically double - peaked , with the first peak missing or very weak , and the appearance of a possible first peak in the @xmath32 100 mev and 0.1 - 0.3 gamma - ray light curves ( signal/@xmath145 2.5@xmath85 ) that is around the phase of the first x - ray peak is intriguing .
but the second gamma - ray peak is not in phase with the second x - ray peak .
as the radiation mechanisms contributing to the emission in x - ray and lat wavebands may differ , and at a given magnetospheric colatitude , the anisotropy of each emission mechanism is both energy - dependent and altitude - dependent , one does not expect the pulse profiles to be achromatic .
therefore , more detailed physical radiation models are required to understand the energy dependence of the lat light curve and to explain the lower energy x - ray light curve . + this evidence against low altitude emission in these pulsars can also be supplemented by constraints of a separate physical origin .
the observation of photons out to beyond 5 gev precludes any dominant action of magnetic pair creation @xmath146 in the emission region at energies below this .
accordingly , the maximum observed energy , @xmath147gev , provides a lower bound to the altitudes of emission , since it must lie below any threshold energy for a super - exponential @xmath0-b pair production turnover .
such constraints have been obtained for @xmath148 detections of the vela pulsar @xcite , and psr j1028 - 5819 @xcite , indicating that the super - gev emission must originate above at least 2.2 and 2.1 stellar radii , respectively . using a standard polar cap model estimate for the minimum emission height of @xmath149 ( e.g. inverting eq .
[ 1 ] of @xcite ) , for a surface polar field strength of @xmath150 g , the psr j1048 - 5832 spin - down parameters ( @xmath151s , @xmath152 ) together with the maximum observed energy @xmath153gev in figure 3 yield @xmath154 .
this drops to @xmath155 if @xmath156gev is deployed when defining this bound ( as was the case for the aforementioned _ fermi _ pulsar detections ) .
similarly , for psr j2229 + 6114 with @xmath157s , @xmath158 and the maximum observed energy @xmath159gev in figure 4 , the altitude constraint is @xmath160 . for either case ,
clearly these bounds preclude emission very near the stellar surface , adding to the advocacy for a slot gap or outer gap acceleration locale for the emission in both pulsars .
although psr j1048@xmath25832 and psr j2229 + 6114 are both vela - like in age and spin characteristics , their light curves and derived emission geometries are quite different .
table [ tab : sumres ] summarises the main quantities measured for both pulsars .
the double - peaked light curve of psr j1048@xmath25832 is nearly identical to that of vela , whereas its derived efficiency is a factor of 10 larger than that of vela @xcite , adopting @xmath161 and @xmath162kpc . on the contrary ,
the @xmath0-ray efficiency of j2229 + 6114 is very similar to that of the vela pulsar , but the pulsar shows a single , large peak similar to psr j0357 + 32 discovered by searching for pulsations at the positions of bright @xmath0-ray sources @xcite .
note that the efficiency of psr j2229 + 6114 would be smaller at the distance of about 1kpc that some authors suggest . with the growing number of detected @xmath0-ray pulsars
, we are beginning to sample a wider variety of emission and viewing geometries , and pulsar ages .
the range of light curve morphologies should allow improved constraints on high - energy emission models and a better understanding of the pulsar magnetospheric structure and acceleration process .
for example , while many young pulsars , like j1048@xmath25832 , show vela - type light curves , a small number are similar to j2229 + 6114 , with a strong p2 component and a weak or absent p1 @xcite .
mapping the angle range over which p1 is missing , especially when viewing angle constraints are available , can help narrow down the high altitude emission zone . a larger pulsar sample also allows a study of evolution of the @xmath0-ray beaming and efficiency with pulsar age : the pulsars seen with _ fermi _
, not including millisecond pulsars , span ages from 10@xmath76 to @xmath163yr @xcite , with hopes to extend that to larger @xmath164 ( lower @xmath38 ) as observations continue .
analysis of the population of pulsars with interpulses @xcite and radio polarization data @xcite have given hints that the magnetic inclination is larger for young pulsars and decreases with age .
thus we may even probe evolution of magnetic alignment during pulsar spindown .
the _ fermi _ lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden .
the parkes radio telescope is part of the australia telescope which is funded by the commonwealth government for operation as a national facility managed by csiro .
we thank our colleagues for their assistance with the radio timing observations .
the lovell telescope is owned and operated by the university of manchester as part of the jodrell bank centre for astrophysics with support from the science and technology facilities council of the united kingdom .
abdo , a. a. , et al .
2008 , science , 322 , 1218 ( cta1 ) abdo , a. a. , et al .
2009a , , 696 , 1084 ( vela ) abdo , a. a. , et al . 2009b , , 700 , 1059 ( psr j2021 + 3651 ) abdo , a. a. , et al .
2009c , , 695 , l72 ( psr j1028@xmath25819 ) abdo , a. a. , et al . 2009d , , 183 , 46 ( bsl ) abdo , a. a. , et al . 2009e , , 699 , l102 ( psr j0205 + 6449 ) abdo , a. a. , et al .
2009f , science , doi : 10.1126/science.1175558 ( blind search pulsars ) abdo , a. a. , et al .
2009 g , , accepted ( arxiv:0904.1018 ) abdo , a. a. , et al . 2009h , astroparticle phys . ,
submitted ( arxiv:0904.2226 ) ( on - orbit calibrations ) abdo , a. a. , et al . 2009i , apj , submitted ( pulsar catalog ) atwood , w. b. , ziegler , m. , johnson , r. p. , & baughman , b. m. 2006 , , 652 , l49 atwood , w. b. , et al .
2009 , , 697 , 1071 baring , m. g. 2004 , adspr , 33 , 552 bhattacharya , d. , akyz , a. , miyagi , t. , samimi , j. , & zych , a. 2003 , , 404 , 163 brisken , w. f. , benson , j. m. , goss , w. m. , & thorsett , s. e. 2002 , , 571 , 906 casandjian , j .-
m . & grenier , i. a. 2008 , a&a , 489 , 849 cordes , j. m. , & lazio , t. j. w. 2002 , arxiv : astro - ph/0207156 daugherty , j. k. , & harding , a. k. 1996 , a&as , 120 , 107 de jager , o. c. , raubenheimer , b. c. , & swanepoel , j. w. h. 1989 , , 221 , 180 dyks , j. , & rudak , b. 2003 , , 598 , 1201 fierro , j. m. 1995 , ph . d. thesis , stanford u. halpern , j. p. , gotthelf , e. v. , leighly , k. m. , & helfand , d. j. 2001a , , 547 , 323 halpern , j. p. , camilo , f. , gotthelf , e. v. , helfand , d. j. , kramer , m. , lyne , a. g. , leighly , k. m. , & eracleous , m. 2001b , , 552 , l125 halpern , j. p. , et al .
2008 , , 688 , l33 harding , a. k. , usov , v. v. , & muslimov , a. g. 2005 , , 622 , 531 hartman , r. c. , et al . 1999 , , 123 , 79 hobbs , g. , lyne , a. g. , kramer , m. , martin , c. e. , & jordan , c. 2004 , , 353 , 1311 hobbs , g. b. , edwards , r. t. , & manchester , r. n. 2006 , mnras , 369 , 655 gonzalez , m. e. , kaspi , v. m. , pivovaroff , m. j. , gaensler , b. m. , 2006 , , 652 , 569 johnston , s. , lyne , a. g. , manchester , r. n. , kniffen , d. a. , damico , n. , lim , j. , & ashworth , m. 1992 , , 255 , 401 johnston , s. , koribalski , b. , weisberg , j. m. , wilson , w. 1996 , , 279 , 661 johnston , s. , karastergiou , a. , & willett , k. 2006 , , 369 , 1916 joncas , g. , & higgs , l. a. 1990 , a&as , 82 , 113 karastergiou , a. , johnston , s. , & manchester , r. n. 2005 , , 359 , 481 kaplan , d. l. , et al .
2005 , , 117 , 643 kaspi , v. m. , lackey , j. r. , mattox , j. , manchester , r. n. , bailes , m. , & pace , r. 2000 , , 528 , 445 kothes , r. , uyaniker , b. , & pineault , s. 2001 , apj , 560 , 236 kramer , m. , et al . 2003 ,
mnras , 342 , 1299 manchester , r. n. 2008 , 40 years of pulsars : millisecond pulsars , magnetars and more , 983 , 584 mattox , j. r. , et al .
1996 , , 461 , 396 muslimov , a. g. , & harding , a. k. 2004 , , 606 , 1143 nolan , p. l. , tompkins , w. f. , grenier , i. a. , & michelson , p. f. 2003 , , 597 , 615 ng , c .- y . &
romani , r. w. 2004 , apj , 601 , 479 ng , c .- y . & romani , r. w. 2008 , apj , 673 , 411 pellizzoni , a. et al . , 2009 ,
apj , 695 , l115 pivovaroff , m. j. , kaspi , v. m. , & gotthelf , e. v. 2000 , , 528 , 436 radhakrishnan , v. , & cooke , d. j. 1969 , , 3 , 225 romani , r. w. 1996 , , 470 , 469 smith , d. a. et al .
2008 , , 492 , 923 standish , e. m. , jpl planetary and lunar ephemerides , de405/le405 , memo iom 312.f-98 - 048 ( 1998 ) strong , a. w. , moskalenko , i. v. , & reimer , o. 2004 , apj , 613 , 962 strong , a. w. , moskalenko , i. v. , reimer , o. , digel , s. , & diehl , r. 2004 , , 422 , l47 swanenburg , b. n. , et al . 1981 , , 243 , l69 tauris , t. m. & manchester , r. n. 1998 , mnras , 298 , 625 thompson , d. j. 2004 , cosmic gamma - ray sources , 304 , 149 yadigaroglu , i .- a . , & romani , r. w. 1995 , , 449 , 211 wang , n. , manchester , r. n. , pace , r. t. , bailes , m. , kaspi , v. m. , stappers , b. w. , & lyne , a. g. 2000 , , 317 , 843 watters , k. p. , romani , r. w. , weltevrede , p. , & johnston , s. 2009 , , 695 , 1289 weltevrede , p. et al .
2009 , pasa , submitted weltevrede , p. et al . 2009b , apj , submitted ( radio polarization of 6 gamma - ray pulsars ) weltevrede , p. , & johnston , s. 2008a , , 387 , 1755 weltevrede , p. , & johnston , s. 2008b , , 391 , 1210 | we report the detection of @xmath0-ray pulsations ( @xmath1 gev ) from psr j2229 + 6114 and psr j1048@xmath25832 , the latter having been detected as a low - significance pulsar by egret .
data in the @xmath0-ray band were acquired by the large area telescope aboard the _ fermi gamma - ray space telescope _ , while the radio rotational ephemerides used to fold the @xmath0-ray light curves were obtained using the green bank telescope , the lovell telescope at jodrell bank , and the parkes telescope .
the two young radio pulsars , located within the error circles of the previously unidentified egret sources 3eg j1048@xmath25840 and 3eg j2227 + 6122 , present spin - down characteristics similar to the vela pulsar .
psr j1048@xmath25832 shows two sharp peaks at phases @xmath3 and @xmath4 relative to the radio pulse confirming the egret light curve , while psr j2229 + 6114 presents a very broad peak at phase @xmath5 .
the @xmath0-ray spectra above 0.1 gev of both pulsars are fit with power laws having exponential cutoffs near 3 gev , leading to integral photon fluxes of @xmath6@xmath7s@xmath8 for psr j1048@xmath25832 and @xmath9@xmath7s@xmath8 for psr j2229 + 6114 .
the first uncertainty is statistical and the second is systematic .
psr j1048@xmath25832 is one of two lat sources which were entangled together as 3eg j1048@xmath25840 .
these detections add to the growing number of young @xmath0-ray pulsars that make up the dominant population of gev @xmath0-ray sources in the galactic plane . |
power corrections to event shape observables in @xmath1 annihilation have been an active field of research in the recent years .
event shapes , as opposed to other inclusive observables , do not have an operator product expansion , so there is no established field theoretic framework to analyse them beyond the perturbative level . on the other hand ,
the experimental data now available cover a wide range of scales and thus could provide an opportunity to test qcd and extract a precise value of @xmath3 .
the state of the art in perturbative calculations of average event shape variables is @xmath4 , i.e. nlo . it turns out that experimental data are not well described by these perturbative expressions , unless explicit power corrections , that may be associated with hadronization , are introduced .
renormalon phenomenology allows to predict the form of the power terms while their magnitude is determined by experimental fits . in the work reported here
@xcite we assume the existence of an abelian like `` dressed skeleton expansion '' in qcd and calculate the single dressed gluon contribution using the dispersive approach .
this way we perform at once all order resummation of perturbative terms which are related to the running coupling ( renormalons ) and parametrization of power corrections .
we discuss the ambiguity between the perturbative sum and the power corrections and show that the resummation is essential in order to extract the correct value of @xmath3 from experimental data .
to demonstrate the method proposed we concentrate on one specific observable , the average thrust , defined by t= , [ t_def ] where @xmath5 runs over all the particles in the final state , @xmath6 are the 3-momenta of the particles and @xmath7 is the thrust axis which is set such that @xmath8 is maximized .
it is useful to define @xmath9 which vanishes in the 2-jet limit .
being collinear and infrared safe , the average thrust can be calculated in perturbative qcd to yield _ ( q^2 ) = [ t_msbar ] where @xmath10 , @xmath11 and the coefficients are @xmath12 and @xmath13 ( see refs . in @xcite ) . using the world average value of @xmath3 , @xmath14 , the nlo perturbative result ( [ t_msbar ] ) turns out to be quite far from experimental data .
this is shown in fig .
[ k_117_pv ] . from the figure it is clear that the nlo correction is quite significant , so
the perturbative series truncated at this order is not very reliable .
furthermore , the renormalization scale dependence is significant .
it follows that higher order corrections that are related to the running coupling can not be ignored .
in order to take running coupling effects into account it is useful @xcite to assume , in analogy with the abelian theory , that there exists a `` dressed skeleton expansion '' .
then , the most important corrections which correspond to a single dressed gluon can be written in the form of a renormalon integral d_0=_0^(k^2/q^2)|a(k^2 ) [ d0 ] where @xmath15 represents a specific `` skeleton effective charge '' , not yet determined in qcd ( in @xcite several schemes were used ) . as opposed to the standard perturbative approch ( [ t_msbar ] ) , by performing the integral ( [ d0 ] ) over all scales one avoids completely renormalization scale dependence .
the integral ( [ d0 ] ) represents a non borel summable power series .
indeed , it involves integration over the coupling constant in the infrared which is ill - defined in perturbation theory due to landau singularities .
the ambiguous integral ( [ d0 ] ) can be defined mathematically , e.g. as a principle value of the borel sum : @xmath16 .
this , however , does not solve the physical problem of infrared renormalons : information on large distances is required to fix the ambiguity .
the perturbative calculation contains some information about the ambiguity : if the leading term in the small momentum expansion of @xmath17 is , the leading infrared renormalon is located at @xmath18 in the borel plane , and a power correction of the form @xmath19 is expected . having no way to handle the problem on the non - perturbative level , it is natural to attempt a fit of the form @xmath20 .
a stronger assumption @xcite is that the `` skeleton coupling '' can be defined on the non - perturbative level down to the infrared .
then the integral ( [ d0 ] ) should give at once the perturbative result plus the correct power term .
since the infrared coupling is not known , it is considered as a non - perturbative parameter . using the cutoff regularization of ( [ d0 ] ) ,
one fits the data with , where d_0__i=__i^2^(k^2/q^2)|a(k^2 ) [ d0_mu_i ] is fully under control in perturbation theory and the normalization of the power term _ _
i = c_n_0^_i^2k^2n|a(k^2 ) [ moment ] is a perturbatively calculable coefficient times a moment of the coupling in the infrared .
since the coupling is assumed to be universal , the magnitude of power corrections can be compared between different observables @xcite .
the generalization of this approach to minkowskian observables such as the thrust was discussed in @xcite . at the level of a single gluon emission it is based on a `` gluon mass '' renormalon integral @xcite @xmath21 where the characteristic function @xmath22 is calculated based on the matrix element for the emission of one massive gluon and @xmath23 is related to the time - like discontinuity of the coupling .
regularizations of the perturbative sum ( [ rapt ] ) with @xmath15 at one or two loops were discussed in @xcite .
we skip it here for brevity .
in order to perform renormalon resummation at the level of a single gluon emission , we calculated the characteristic function for the thrust @xcite . then @xmath24 is evaluated taking either the one or two loop coupling . since @xmath24 does not exhaust the full nlo correction of eq .
( [ t_msbar ] ) , we add an explicit nlo correction , @xmath25 , where @xmath26 corresponds to ( [ t_msbar ] ) and @xmath27 is the piece included in @xmath24 .
it was shown in @xcite that the abelian part of @xmath27 almost coincides with that of @xmath26 in spite of the non - inclusive nature of the thrust .
this is crucial for the applicability of the current resummation approach . finally , we add an explicit power correction @xmath28 where the power @xmath29 is determined from the small @xmath30 expansion of @xmath22 .
thus , using the pv regularization of @xmath24 , we fit the data with = + .
[ inc_lambda ] the resummation as well as the fit results are presented in fig .
[ k_117_pv ] for the world average value of @xmath3 .
the resummation by itself is quite close to the data .
note also the stability of the result as @xmath15 is promoted from one to two loops .
next , we fit also the value of @xmath3 .
the best fit @xmath31 , obtained with @xmath32 and @xmath33 , is shown in fig .
[ pv_cutoff_110 ] .
it is interesting to compare the resummation to a truncated expansion in @xmath35 .
expanding @xmath24 with @xmath15 at 1-loop we obtain a series of the form @xmath36 .
the coefficients @xmath37 can be considered as predictions for the perturbative coefficients @xmath38 , provided the coupling @xmath15 is close to the correct `` skeleton coupling '' , and provided that the `` leading skeleton '' @xmath24 is indeed dominant already in the sub - asymptotic regime .
the significance of the nnlo and further sub - leading corrections which correspond to the dissociation of the emitted gluon is demonstrated in fig .
the figure shows @xmath39 where @xmath40 is the order of truncation for @xmath41 through @xmath42 . at these orders the series is still convergent .
we conclude that a major part of the discrepancy between the nlo result and experiment is due to neglecting this particular class of higher order corrections .
next consider a fit to experimental data based on the truncated expansion of @xmath24 ( [ trunc_fit ] ) : .
the fit results are listed in table [ trunc_fit_tab ] .
[ trunc_fit_tab ] @xmath43 the quality of the fit is roughly the same in all cases .
we see that the resummation is absolutely necessary in order to extract a reliable value of @xmath3 .
putting an infrared cutoff @xmath44 on the space - like momentum ( [ d0_mu_i ] ) we separate at once the perturbative and non - perturbative regimes as well as large and small distances . the cutoff regularized sum , @xmath45 for is shown in fig . [ pv_cutoff_110 ] together with the pv regularization .
for the lower @xmath46 values the difference between the two regularizations is large .
this means that a large contribution to @xmath47 comes from momentum scales below @xmath48 , where the coupling is not controlled by perturbation theory . finally fitting the data = + [ inc_lambda_cutoff ] we get the same result as with the pv regularization ( [ inc_lambda ] ) .
the only difference is in the required power correction @xmath49 .
this is general : the two regularizations differ just by ( calculable ) infrared power corrections in our case @xmath2 and are therefore equivalent once the appropriate power terms are included .
the assumption of a `` skeleton expansion '' implies that resummation of perturbation theory and parametrization of power corrections must be performed together . for the thrust renormalon resummation is significant and closes most of the gap between the standard perturbative result ( nlo ) and experiment .
the resummation is crucial to extract the correct value of @xmath3 .
* e. de rafael * _ what is the physical meaning , in qcd , of the scale of the @xmath2 correction ?
what other processes are sensitive to these corrections ? _
+ _ the first question is basically an open one .
deeper understanding could hopefully be gained once renormalon phenomenology is supported by more rigorous field theoretic methods . in the infrared finite coupling approach the @xmath2 correction is understood as a moment ( [ moment ] ) of a universal infrared finite coupling .
this allows comparison between observables for example the c parameter is sensitive to similar @xmath2 corrections . in the framework of shape - functions @xcite , a relation with the energy - momentum tensor was suggested .
_ 9 e. gardi and g. grunberg , [ hep - ph/9908458 ] .
brodsky , g.p .
lepage and p.b .
mackenzie , _ phys .
* d28 * ( 1983 ) 228 . yu.l .
dokshitzer , g. marchesini and b.r .
webber , _ nucl .
* b469 * ( 1996 ) 93 .
m. beneke and v.m .
braun , _ phys .
_ * b348 * ( 1995 ) 513 .
korchemsky and g. sterman , [ hep - ph/9902341 ] . | infrared power corrections for the average thrust @xmath0 in @xmath1 annihilation are analyzed in the framework of renormalon resummation , motivated by analogy with the skeleton expansion in qed and the blm approach . performing the `` massive gluon '' renormalon
integral a renormalization scheme invariant result is obtained .
we find that a major part of the discrepancy between the known next - to - leading order ( nlo ) calculation and experiment can be explained by resummation of higher order perturbative terms .
this fact does not preclude the infrared finite coupling interpretation with a substantial @xmath2 power term .
fitting the regularized perturbative sum plus a @xmath2 term to experimental data yields .
loop loop |
in the last few years the astronomical community has moved much closer to unveiling the nature of grbs .
the discovery of afterglows and identification of host galaxies for several bursts clearly links grbs to some type of stellar events .
yet the nature of these events is unknown , and we still do not know what the grb central engines are .
observations of grb host galaxies and precise locations of grbs within hosts provide a tool to test some of the possible central engine models . in this paper
we discuss the consistency between the current observations and the results of binary population synthesis .
the population synthesis code used here is described in @xcite .
one of the most important parameter deterimining the properties of the populations of binaries is the kick velocity a newly formed compact object receives at birth .
however , several studies @xcite indicate that the distribution of kick velocities consists of two components : a low velocity with the width of approximately @xmath0kms@xmath1 , and a high velocity with the characteristic velocity around @xmath2kms@xmath1 .
about 80% of the kicks are drawn from the first component .
it is also known that the production rate of compact object binaries falls off exponentially with increasing kick velocity , see fig . 1 in @xcite .
thus the population of compact object binaries will be dominated by the objects formed in the systems that received the kicks drawn from the low velocity component of the distribution . in the following
we will consider the properties of the compact object binaries for the case when the kick velocity is drawn from a gaussian distribution with the width @xmath0kms@xmath1 .
little is known a priori about the masses and gravtational potentials of host galaxies where grb progenitrs reside . therefore to find the expected distribtion of merger sites around galaxies we consider two extreme cases : propogation in a potential of a large galaxy like the milky way and propagation in the empty space @xcite .
in figure 1 we present the distribution of center of mass velocities gained by systems in the supernova explosions in the galactic potantial and binary lifetimes ( the time binary takes to evolve from zams to final merger of two components ) for four types of compact object binaries . for
the case of propagation in a potential of a massive galaxy only a fraction of ns - ns binaries will be able to escape from their host galaxies .
the bh - ns binaries tend to stay in the galaxy . here
, we have assumed that the kick velocity does not depend on the nature of the compact object formed in the supernova .
however , it has been argued that the kicks back holes receive should be smaller than those of the neutron stars .
this is discussed in more detail in @xcite .
the helium star mergers stay in the host galaxies , while some of white dwarf black hole mergers have a chance of taking place outside the host , provided that the escape velocity from a given galaxy is not too large . in the case of propagation in empty space
quite a large number of binaries of any type will be able to escpae from their birthplace . in figure 2
we present the cumulative distributions of the distances ( projected on the sky ) between mergers sites and the host galaxies . in the case of propagation in empty space (
left panel of figure 2 ) most of the mergers take place far outside the host . in the case of propagation in the potential of a massive galaxy ( right panel of figure 2 ) black hole neutron star mergers and helium star mergers take place inside the host , and only a small , but not negligible fraction of double neutron stars and black hole white dwarf mergers happen outside the host .
we have learned at this conference @xcite that the afterglow locations coincide with galaxies , and that typically there are intense star forming processes in these galaxies . in the compact object merger model of grbs one
has to take into account the fact there are significant delays between the stellar formation and the time of merger .
this delays consists of the stellar evolutionary time leading to supernavae explosions and formation of the compact onbject binary and the evolution of the compact object binary due to gravitational wave energy loss .
the distribution of the delay times is rather wide and varies for different types of binaries @xcite , see figure 1 . in the case of
helium star mergers these delays can be as short as a few million years . assuming that grbs are related with ns - ns , bh - ns , or bh - wd mergers we do not expect any correlation between the grb sites and star formation because the star formation processes could have ceased by the time the merger happens .
thus , within this model grb rate should be proportional to the luminous mass in the universe . as most of the luminous mass is concentrated in massive galaxies we expect to find grbs within such galaxies
. however , this should be typical galaxies and we tend to find grb hosts in small , star forming ones @xcite .
let us now consider the case when the delays between the stellar formation and the merger events are shorter than the star forming episode itself .
naturally , a correlation between the grb sites and star forming galaxies exists .
however , the observed host galaxies in this case are typically small .
thus , as shown above , a significant fraction of the mergers should take place outside the host galaxies and we should be finding grbs with no underlying host galaxies .
one can also argue that the grbs that take place outside the host galaxies do not produce significant afterglows , because of low density of the ambient medium and therefore are not followed by afterglows .
this would make a strong selection effect against detecting grbs happening outside of the hosts . at this conference
we have heard however , that the beppo sax data are consistent with all grbs having x - ray afterglows @xcite .
this means that all grbs have afterglows and that grbs with no afterglows do not exist , at least within the sample observable by beppo sax .
the reasoning presented above strongly argues against the compact object merger model of grbs .
however , we must remember that afterglows have been detected only from the long bursts , and we do not know if short bursts also produce afterglows and what are the locations in relation to host galaxies of short bursts .
moreover , numerical models of compact object coalescences @xcite agree with analytical estimates and show that the timescales of these events can not be stretched beyond a fraction of a second . yet
the long bursts have a median duration of approximately @xmath3s .
thus we conclude that the compact object merger model appears to be inconsistent with the observations of afterglows , and their locations within the host galaxies .
long bursts are therefore most probably not connected with compact object mergers .
however , it is quite likely that we will find that the short bursts are connected with mergers of compact objects . | we use a stellar binary population synthesis code to find the lifetimes and velocities of several types of possible grb progenitors : double neutron stars , black hole neutron stars , black hole white dwarfs , helium star mergers . assuming that they are born in different types of galaxies we compute their spatial distribution and compare it with the observed locations of grb afterglows within their hosts .
we discuss constraints on the compact object merger model of grbs imposed by this comparison and find that the observations of afterglows and their host galaxies appear inconsistent with the grb compact object merger model . |
the penrose cosmic censorship states that : if singularities predicted by general relativity @xcite occur , they must be dressed by event horizons @xcite .
hence , the search of the regular black hole solutions has been a focus of research interests lately @xcite .
in fact , the study of global regularity of black hole solutions is important for the understanding of the final state of the gravitational collapse from some initially regular configurations . earlier work on regular black hole models can be found in references @xcite .
these regular solutions are referred to as `` bardeen black holes '' @xcite .
in addition , regular black hole solutions to einstein equations with various physical sources were reported in @xcite . the other approaches to avoid the singularities are the search of some more generalized theories , for example , the string / m theory @xcite .
the einstein equation reads @xmath0 where @xmath1 is the einstein tensor and @xmath2 is the matter stress - energy tensor .
for example , @xmath3 is the contribution from a cosmological constant @xmath4 . in the absence of the @xmath2 associated with the other matter
, @xmath4 must be a constant .
this follows from the bianchi identity : @xmath5 .
there have been activities trying to treat the cosmological constant @xmath4 as a dynamical quantity @xcite .
the vacuum stress - energy tensor has the form @xmath6 in the semi - classical approach of the quantum field theory in curved space . in this approach ,
the cosmological term becomes @xmath7 which has been studied for more than two decades @xcite .
note that the de sitter - schwarzschild solution found in ref .
@xcite is a black hole solution with a de sitter core near @xmath8 .
we will generalize this solution to the charged regular solution by assuming that the matter coupled to the gravitational system carries charge .
we assume that the charged material has a uniform charge / mass ratio all over .
therefore , the @xmath9-dependence of the charged density @xmath10 is the same as the @xmath9-dependence of the mass density @xmath11 for the charged material .
we will show that a regular charged black hole solution exists with a charged de sitter core .
we will also study the properties of this regular charged solution and its possible impact .
in the static and spherically symmetric space , the line element can be written in the form @xcite
@xmath12 dt^2 + \exp[2b(r ) ] dr^2 + r^2 d\omega\ ] ] where @xmath13 is the solid angle .
one can write @xmath14 = 1- 2n(r)/r$ ] and show that the einstein equation @xmath15 reads @xmath16 it is known that the energy momentum tensor of the static and spherically symmetric star takes the following form @xmath17 with @xmath18 signifies the pressure contribution . for simplicity we will work on the case where @xmath19 which is classified as the type [ ( ii),(ii ) ] field according to the classification of serge @xcite .
the @xmath20 and @xmath21 components of the einstein equation give immediately that @xmath22 once we choose the proportional constant to be zero .
in addition , the conservation of energy momentum @xmath23 requires that @xmath24 this equation tells us what the pressure term @xmath18 should look like with a given energy density @xmath25 in order to stabilize the space induced by the serge s type [ ( ii),(ii ) ] field we are interested .
for example , @xmath26 if @xmath27 .
in addition , one has @xmath28 for constant @xmath29 , @xmath30 and @xmath4 according to equation ( [ n - eq ] ) .
note that this is the reissner - nordstrom solution with a cosmological constant term .
note that the @xmath31 gauge field will provide a lagrangian of the form @xmath32 this will contribute a traceless energy momentum tensor of the form @xmath33 to the einstein equation .
in addition , the @xmath34 electric charge conservation @xmath35 with a static source @xmath36 gives the system a static coulomb potential @xmath37 in the coulomb s gauge .
one can show that the electric field is @xmath38 for this spherically symmetric source . here
@xmath39 is the total charge surrounded by the @xmath9-sphere .
hence the energy momentum tensor for this charged material can be shown to be @xmath40 here we purposely denote the charge density as @xmath41 because the mass density happens to be the energy density of the matter sector while the energy density of the photon sector @xmath42 is not the same as the charge density @xmath41 .
note also that one should have @xmath43 if the photon energy momentum tensor is conserved , @xmath44 , by itself .
but , this is true only if the total charge @xmath30 is a constant everywhere .
hence , the conservation of energy momentum tensor @xmath45 is known to be in - consistent with the @xmath46-dependent static electric field given by @xmath47 for any @xmath9-dependent @xmath48 .
in fact , the conservation law requires an @xmath49 energy density because @xmath50 for a the coulomb s electric field with a constant @xmath30 .
this easy to verify since the operator @xmath51 is nothing more than a power counting operator .
hence the conservation equation @xmath52 simply implies that @xmath53 .
this is a problem one has to deal with in working with the charged solutions for a regular distribution of charged material .
in fact , the resolution is to assign an additional pressure contribution to the matter sector .
indeed , one can show that the total energy momentum conservation remain valid if the total pressure @xmath18 is given by @xmath54\ ] ] for the @xmath9-dependent @xmath46 given earlier . here
@xmath55 is the matter pressure in the absence of the @xmath9-dependent @xmath30 . in other word , the system remains consistent if the matter pressure @xmath56 offers an extra contribution in order to counteract the contribution due to the @xmath46 term . note that we only show that a stable static system requires an additional pressure for stability .
stable charged stars certainly can exit and this extra pressure simply signifies the contribution from a complicated interaction among the charged material .
let s imagine a charged material distribution of the following form @xmath57 / 4 \pi , \\
\label{cme } d_e & = & 3q_0 \exp [ -r^3 ] / 4 \pi.\end{aligned}\ ] ] one can show that the total mass and total charge measured at a distance @xmath9 read @xmath58 \,)$ ] and @xmath59 \,)$ ] for this charged material .
in fact , one can show that the energy of the whole system is @xmath60 / 4 \pi + [ q_0 ^ 2/8 \pi r^4 ] ( 1- \exp [ -r^3 ] \,)^2
. \label{pm}\end{aligned}\ ] ] note that @xmath61 and @xmath62 denote the total mass and total charge measured at spatial infinity .
one has assumed that each charged matter @xmath63 carries mass @xmath64 and a fixed unit of charge @xmath65 such that @xmath66 .
here @xmath67 denotes the total number of the molecular and/or charged material @xmath63 .
one can show that the matter pressure of this system is given by @xmath68 / 4 \pi - 9m_0 r^3 \exp [ -r^3 ] / 8 \pi + 3q_0 ^ 2 \exp [ -r^3 ] ( 1- \exp [ -r^3 ] ) / 8 \pi r \label{cmprm}\ ] ] with the last term signifies the counter term from the charged sector .
one can imagine that charged material carrying mass quanta and charge quanta tries to form a regular black hole with a charged de sitter ( cds ) core .
if the uncharged matter can form a regular de sitter - schwarzschild black hole , the charged material should also be able to form a charged regular black hole .
hence we will try to study the possibility of finding such kind of regular black hole solution .
note that we have changed the scale @xmath9 to be dimensionless such that @xmath9 should be @xmath69 for a physical scale @xmath70 typically of the size of radius of the event horizon . in this approach
, the charged material is bounded mostly inside the @xmath70-sphere .
note that the weak energy condition , @xmath71 , for any time - like vector @xmath72 is violated as @xmath73 for this solution .
indeed , one can show that @xmath74/2 - 3q_0 ^ 2\exp [ -r^3 ] ( 1- \exp [ -r^3])/2r \ge 0\ ] ] therefore , the weak energy condition also requires @xmath75 for all @xmath76 .
this inequality will break down for @xmath77 .
hence the existence of such solution will require the existence of exotic matter @xcite .
one way out is to assume the charge distribution grows as @xmath78 $ ] , namely , @xmath79 ) $ ] for some @xmath80 .
one can show that the inequality @xmath81 is needed for a regular solution which demand all scale - invariant curvature - square terms like @xmath82 be finite everywhere .
hence one can show that the weak energy condition reads @xmath74/2-lq_0 ^ 2\exp [ -r^l ] ( 1- \exp
[ -r^l])/2r \ge 0\ ] ] therefore , the weak energy condition becomes @xmath83 when @xmath76 .
this inequality can be made valid for all @xmath76 if @xmath84 and @xmath85 .
note that this is not the necessary condition for the weak energy condition .
therefore , smaller charge accumulated around the matter distribution can save the weak energy condition from breaking - down .
note that the charged black hole of this type can be shown to give the following metric solution @xmath86 ) /r
+ q_0 ^ 2 ( 1 - \exp [ -r^l ] ) ^2 /r^2 - [ 2l q_0 ^ 2/r ] \int_0^r r'^{l-2 } dr ' \exp [ -r'^l ] ( 1 - \exp [ -r^l ] ) .
\label{grr}\ ] ] we will focus on the case where @xmath84 for the moment . note that the last term can be integrated to give @xmath87 \
{ \gamma[7/4]/3 - \gamma [ 3/4 , r^4 ] /4- ( \gamma[7/4 ] - \gamma [ 3/4 , 2r^4])/2^{3/4 } \}\ ] ] described by the di - gamma function when @xmath84 .
note that the last term can be shown to give @xmath88 if @xmath76 .
it is @xmath89 if @xmath76 for arbitrary @xmath90 .
in fact , if the horizon is located in region @xmath76 , the metric @xmath91 reads @xmath92 .
this equation is different from the reissner - nordstrom solution by the sign of the last term related to @xmath93 .
hence there always exist two distinct event horizons for this regular charged black hole if the horizon exists deep inside the @xmath70-core , i.e. @xmath94 .
but this is not the correct picture for a black hole .
matter should reside mostly inside the black event horizon where @xmath95 diverges .
hence one expects that the radius of the event horizons should be greater than the typical size of the @xmath70-core .
therefore , the above naive picture does not hold for the region where @xmath96 .
in fact , numerical analysis indicates that there is an event horizon and a cauchy horizon if @xmath97 in our normalized unit @xmath98 .
numerical results are plotted in fig .
i. ( 25,100 ) ( -60,20 ) figure i is plotted only for reference with the total charge set equal to @xmath99 and mass @xmath61 set as @xmath100 .
the lower thick curve denotes the @xmath91-@xmath9 curve for the case @xmath101 with the last digamma term in eq .
( [ grr ] ) .
the lower thin curve goes for the case @xmath102 with the last digamma term in eq .
( [ grr ] ) .
note that the lower thin curve indicates the case where extreme charged black hole exists .
these are plotted with @xmath91 given by eq .
( [ grr ] ) .
in addition , the upper thin / thick curves are the curves plotted only with @xmath91 given by the first three terms in eq .
( [ grr ] ) .
this result indicates that the contribution of the di - gamma functions can not be ignored .
note also that all curves approach @xmath103 at spatial infinity .
it is known that the event horizon disappears for the reissner - nordstrom solution when @xmath104 .
the extreme black hole is , however , not present at the condition @xmath105 in this case .
in fact , the mass / charge distribution introduced here deforms the extreme condition into a very complicated condition .
we are , unfortunately , unable to solve algebraically for the exact expression of the @xmath106-@xmath61 relation for the condition of the extreme black hole .
the numerical study shows , however , that the physical behavior of the black hole horizons is similar to the reissner - nordstrom black hole .
event horizon will disappear under a somewhat complicated @xmath106-@xmath61 relation .
note also that one can show that all curvature tensor @xmath107 for the regular charged black hole solution we just obtained .
therefore , this charged solution is indeed regular at @xmath8 since all @xmath82-invarinats are finite at @xmath8 .
in summary , we have shown that there is a class of regular charged black hole with a charged de sitter core similar to the de sitter - schwarzschild black hole for the type [ ( 2,2 ) ] charged matter .
the associated conservation law of energy momentum tensor is quite different from the uncharged counterpart .
one can show that the energy momentum tensor associated with the static and spherically symmetric photon field is not conserved by itself .
the resolution is to assign an extra photon contribution to the pressure term associated with the matter field @xmath108 .
possible impact and the numerical solutions of the charged black hole are also shown in this paper .
the structure of this cds solution could be important to the physics of the gravitationally collapsing material .
in addition , a more general form of the matter distribution can also be shown to establish the stability of these models @xcite .
this is the first regular exact solution with a charged de sitter core in general relativity based on a physical distribution of charged material @xcite
. it should be important to the understanding of the star evolution if the de sitter - schwarzschild solution is realized in nature .
it is known that the case when @xmath109 will lead to the demand of exotic matter that violates the weak energy condition when @xmath9 is small .
this problem can be resolved by assuming a smaller charge distribution function , e.g. assuming @xmath84 .
we show explicitly that the weak energy condition remains valid for the choice of @xmath84 provided that @xmath61 is much larger than @xmath93 . for simplicity
, we only discuss the case where @xmath84 in most of this paper .
the generalization to arbitrary @xmath90 is straightforward .
no - hair theorem states that all static black holes are characterized uniquely by mass and charge , and have the reissner - nordstrom form .
similar to the regular solution obtained in ref .
@xcite , our solution also appears to violate the no - hair theorem .
in fact , the existence of this sort of regular solution requires a mass and charge distribution of the form given by eq.s ( [ cmr ] , [ cme ] ) along with a pressure given by eq.s ( [ cmprm ] ) and ( [ cmpre ] ) .
this is quite different from the usual vacuum solution we have before .
for example , the behavior of a test particle in - falling to the black hole will require additional knowledge about the constituent of the static @xmath110 .
numerical result also indicates that the size of event horizon is of the same order of the size that most of the matter resides , i.e. @xmath111 while matter ( charge ) density decreases exponentially according to @xmath112\ : ( \exp[-r^4])$ ] .
in addition , the regular solution we just obtained approaches the r - n solution asymptotically . therefore , the only measurable quantities in asymptotic region are @xmath106 and @xmath61 .
hence one can fairly say that the no - hair theorem holds softly for this solution too .
this is because all information beyond the event horizon is not reachable from exterior observer .
even it is not clear how one can arrange mass and charge the way we want in the core region , it is still a heuristic analysis to study regular models similar to the one studied in this paper .
99 hawking , s.w . ,
ellis , g.f .
, _ the large scale structure of space time _ ( cambridge univ . press 1973 ) .
senovilla , _ gen .
_ , * 30 * , 701 ( 1998 ) . r.m .
gr - qc/9710068 ; j. jhingan , g. magli , gr - qc/9903103 ; e. ayon - beato , a. garcfa , phys.rev.lett . 80 ( 1998 ) 5056;phys.lett .
b464 ( 1999 ) 25 ; c. lechner , s. husa , p. c. aichelburg , phys.rev .
d62 ( 2000 ) 044047 ; m. cataldo , a. garcia , phys.rev . d61 ( 2000 ) 084003 ; k. a. bronnikov , gr - qc/0006014 ; j. bardeen , in proceedings of gr5 , tiflis , u.s.s.r .
beato , _ asymptotic behavior of scalar fields coupled to gravity _ , graduate dissertation , faculty of physics , havana univ .
borde , a. , _ phys . rev .
_ , * d50 * , 3392 ( 1994 ) .
barrabs , c. , frolov , v.p . , _ phys
_ , * d53 * , 3215 ( 1996 ) .
mars , m. , martn
prats , m.m . , senovilla , j.m.m .
, _ class .
_ , * 13 * , l51 ( 1996 ) .
cabo , a. , ayn
beato , e. , _ int . j. mod
_ , * a14 * , 2013 ( 1999 ) .
borde , a. , _ phys .
_ , * d55 * , 7615 ( 1997 ) .
beato , e. , garca , a. , _ phys .
_ , * 80 * , 5056 ( 1998 ) .
magli , g. , `` physically valid black hole interior models , '' gr - qc/9706083 ( 1997 ) .
tseytlin , a.a .
b363 * , 223 ( 1995 ) . n. a. bahcall , j. p. ostriker , s. perlmutter , p. j. steinhardt , science * 284 * , ( 1999 ) 1481 .
l. smolin , nucl . phys . * b160 * ( 1979 ) 253 ; a. zee , phys . rev . lett .
* 42 * , 417 ( 1979 ) ; * 44 * , 703 ( 1980 ) .
w.f.kao , phys .
149a , ( 1990 ) 76 ; phys . lett . 154a ( 1991 ) 1 ; phys .
d46 ( 1992 ) 5421 ; w.f .
kao , shi - yuun lin and tzuu - kang chi , phys .
d53 ( 1996 ) 1955 ; w.f .
kao , w.b .
dai , shang yuun wang , tzuu - kang chi and shi - yuun lin , rev .
d53 , ( 1996 ) 2244 ; w.f .
kao , phys .
d61(2000 ) 047501 ; d61(2000)044004 ; hep - th/0003206,d62 ( 2000)087301 ; hep - th/0003153,d62(2000 ) 084009 ; f.s .
accetta , d.j .
zoller and m.s .
turner , phys rev .
d * 31 * 3046 ( 1985 ) ; a.s .
goncharov , a.d .
linde and v.f .
mukhanov , int . j. mod
a2 * 561 ( 1987 ) ; s. l. adler , rev .
phys . * 54 * , 729 ( 1982 ) .
s. weinberg , rev .
phys . * 61 * , 1 ( 1989 ) .
j. m. overduin and f. i. cooperstock , phys . rev .
* d58 * , 043506 ( 1998 ) .
e. b. gliner , sov .
jetp * 22 * , 378 ( 1966 ) .
b. zeldovich , sov .
* 6 * , 883 ( 1968 ) .
a. d. linde , sov .
* 19 * , 183 ( 1974 ) .
e. b. gliner , i. g. dymnikova , sov .
* 1 * , 93 ( 1975 ) .
a. d.dolgov , in _ the very early universe _ ,
g. w. gibbons , s. w. hawking , s. t. c. siclos ( cambridge university press , cambridge , england 1983 ) , p. 449 .
a. z. petrov , _
einstein spaces _ , pergamon press , oxford , 1969
. i. g. dymnikova , gen . rel
* 24 * , 235 ( 1992 ) ; int .
phys . * d5 * , 529 ( 1996 ) ; gravitation and cosmology * * 5 , 15 ( 1999);phys .
b472 ( 2000 ) 33 ; r. c. tolman , _ relativity , thermodynamics and cosmology _ , clarendon press , oxford , 1969 . w.f .
kao , in preparation .
morris , k.s .
thorne , _ wormholes in space - time and their use for interstellar travel : a tool for teaching general relativity _ , am .
j. phys.*56*:395 - 412 ( 1988 ) ; m.s .
morris , k.s .
thorne , u. yurtsever , _
wormholes , time machines and the weak energy condition _ , phys .
* 61*:1446 - 1449 ( 1988 ) ; | we show that there is a class of regular charged black hole with a charged de sitter core similar to the de sitter - schwarzschild black hole .
one can show that the total energy momentum tensor for the static and spherically symmetric photon field can be made conserved even the coulomb like energy momentum tensor for photon is not conserved alone .
possible impact and the numerical solutions of the charged black hole will be shown in this paper . |
one of the hurdles for identifying clinically meaningful patterns in medical data is the fact that much of that data is sparsely , irregularly , and asynchronously observed , rendering it a poor substrate for many pattern recognition algorithms .
a large class of this problematic data in medical records is time - stamped categorical data such as billing codes .
an icd-9 billing code , for example , with value 714.0 ( rheumatoid arthritis ) gets attached to a patient record every time the patient makes contact with the healthcare system for a problem or activity related to her arthritis .
this could be an outpatient doctor visit , a laboratory test , a physical therapy visit , the discharge event of an inpatient stay , or any other billable event .
these events occur at times that are in general independent from events for other conditions .
we would like to learn things from the patterns of these clinical contact events both within and between diseases , but their ( often sparse and ) irregular nature makes it difficult to apply standard learning algorithms to them . to abstract away this problem , we consider the data as streams of events , one stream per code or other categorical label .
we model each stream as a modulated renewal process and use the process s modulation function as the abstract representation of the continuous , longitudinal intensity of the patient s contact with the healthcare system for a particular problem at any point in time .
the resulting inference problem is to estimate a probability density over the renewal process parameters and intensity functions given the raw event data .
the practical utility of using a continuous function density to couple standard learning algorithms to sparse and irregularly observed continuous variables has been previously demonstrated @xcite .
unfortunately , the method of inferring such densities for continuous variables is not applicable to categorical variables .
this paper presents a method that achieves the inference for categorical variables .
our method models the log intensity functions non - parametrically as gaussian processes , and uses markov chain monte carlo ( mcmc ) to infer a posterior distribution over intensity functions and model parameters given the events ( section [ sec : mrp ] ) .
there are several existing approaches to making this inference ( section [ sec : prior - work ] ) , but all of the approaches we found have either flexibility or scalability problems with our clinical data .
for example , clinical event streams can be bursty , and some existing methods are unable to adapt to or adequately represent this . in this paper
we demonstrate using synthetic data that our approach has accuracy , efficiency , and flexibility advantages over the best existing method ( section [ sec : experiment - synthetic ] ) .
we further demonstrate these properties using synthetic data that mimics our clinical data , under conditions that no existing method that we know of is able to satisfactorily operate ( section [ sec : experiment - synthetic ] ) .
finally , we use our method to infer continuous abstractions over real clinical data ( section [ sec : experiment - clinical ] ) .
a renewal process models random events by assuming that the interevent intervals are independent and identically distributed ( iid ) .
a modulated renewal process model drops the iid assumption and adds a longitudinal intensity function that modulates the event rate with respect to time .
we consider a set of event times @xmath0 to form an event stream that can be modeled by a modulated renewal process . for this work
we choose a modulated gamma process @xcite , which models the times @xmath1 as @xmath2 where @xmath3 is the gamma function , @xmath4 is the shape parameter , @xmath5 is the modulating intensity function , and @xmath6 .
equation is a generalization of the homogeneous gamma process @xmath7 , which models the interevent intervals @xmath8 as positive iid random variables : @xmath9 where @xmath10 takes the place of a now - constant @xmath11 , and can be thought of as the time scale of event arrivals .
the intuition behind is that the function @xmath12 warps the event times @xmath13 into a new space where their interevent intervals become draws from the homogeneous gamma process of .
that is , the warped intervals @xmath14 are modeled by @xmath15 . for our purposes ,
a gamma process is better than the simpler and more common poisson process because a gamma process allows us to model the relationship between neighboring events , instead of assuming them to be independent or memoryless .
specifically , parameterizing @xmath16 models a bursty process , @xmath17 models a more regular or refractory , and @xmath18 produces the memoryless poisson process .
clinical event streams can behave anywhere from highly bursty to highly regular .
we model the log intensity function @xmath19 as a draw from a gaussian process prior with zero mean and the squared exponential covariance function @xmath20 where @xmath21 sets the magnitude scale and @xmath22 sets the time scale of the gaussian process .
we choose the squared exponential because of its smoothness guarantees that are relied upon by our inference algorithm , but other covariance functions could be used . in our application
the observation period generally starts at @xmath23 , and ends at @xmath24 , and no events occur at these endpoints .
consequently , we must add terms to to account for these partially observed intervals . for efficiency in inference
, we estimate the probabilities of these intervals by assuming that @xmath25 and @xmath26 are drawn from a homogeneous @xmath27 process in the warped space . the probability of the leading interval @xmath28 is then approximated by @xmath29 , which is equivalent to @xmath30 .
the trailing interval is treated similarly .
our full generative model is as follows : 1 .
[ item : hyperpriors ] @xmath31 + @xmath32 + @xmath33 + @xmath34
[ item : gp]@xmath35 using 3 .
[ item : transform]@xmath36 4 .
[ item : integral ] @xmath6 5 .
@xmath30 ; @xmath37 6 .
[ item : inversion ] @xmath38 step [ item : hyperpriors ] places a prior on @xmath22 that prefers smaller values , and uninformative priors on @xmath39 and @xmath21 .
we set @xmath34 to avoid an identifiability problem .
( @xcite set @xmath40 to avoid this problem .
while that setting has some desirable properties , we ve found that setting @xmath34 avoids more degenerate solutions at inference time . )
given a set of times @xmath1 , we use mcmc to simultaneously infer posterior distributions over the intensity function @xmath41 and the parameters @xmath39 , @xmath21 , and @xmath22 ( algorithm [ alg : inference ] ) , where for simplicity we denote @xmath42 . on each round
we first use slice sampling with surrogate data ( * ? ? ?
* code publicly available ) to compute new draws of @xmath43 , @xmath21 , and @xmath44 using as the likelihood function ( with additional factors for the incomplete intervals at the ends ) .
we then sample the gamma shape parameter @xmath39 with metropolis - hastings moves under the same likelihood function .
one challenge of this direct inference is that it requires integrating @xmath45 , which is difficult because @xmath46 does not have an explicit expression . under certain conditions ,
the integral of a gaussian process has a closed form @xcite , but we know of no closed form for the integral of a log gaussian process .
instead , we compute the integral numerically , relying on the smoothness guarantees provided by the covariance function to provide high accuracy .
the efficiency bottleneck of the update is the @xmath47 complexity of updating the gaussian process @xmath48 at @xmath49 locations , due to a matrix inversion .
naively , we would compute @xmath48 at all @xmath50 of the observed @xmath13 , with additional points as needed for accuracy of the integral . to improve efficiency ,
we do not directly update @xmath48 at the @xmath13 , but instead at @xmath51 uniformly spaced points , where . we then interpolate the values @xmath52 from the values of @xmath53 as needed .
we set the number of points @xmath51 by the accuracy required for the integral .
this is driven by our estimate of the smallest likely gaussian process time scale @xmath54 , at which truncate the prior on @xmath22 to guarantee @xmath55 .
the efficiency of the resulting update is @xmath56 , with @xmath51 depending only on the ratio @xmath57 .
it helps that the time constant driving @xmath51 is the scale of changes in the modulating intensity function instead of the scale of interevent intervals , which is usually much smaller . in practice , we ve found @xmath58 to work well for nearly all of our medical data examples , regardless of the observation time span , resulting in an update that is linear in the number of observed points .
additionally , the regular spacing in @xmath59 means that its covariance matrix generated by is a symmetric positive definite toeplitz matrix , which can be inverted or solved in a compact representation as fast as @xmath60 @xcite .
we did not include this extra efficiency in our implementation , however .
event times @xmath1 , regular grid @xmath59 , current function @xmath48 and parameters @xmath21 , @xmath22 , and @xmath39 updated @xmath48 , @xmath61 , @xmath62 , @xmath21 , @xmath22 , and @xmath39 with likelihood @xmath63 update @xmath64 , using slice sampling compute @xmath52 by smooth interpolation of @xmath53 @xmath65 compute @xmath66 from @xmath67 numerically compute @xmath63 = @xmath68 using update @xmath39 and @xmath63 with metropolis - hastings and
.33 .33 .33 .33 .33 .33 there is a growing literature on finding patterns among clinical variables such as laboratory tests that have both a timestamp and a value ( see * ? ? ?
* and its references for examples ) , but we are not aware of any existing work exploring unsupervised , data - driven abstractions of the purely time - domain clinical event streams that we address here .
there is much prior work on methods similar to ours that infer intensity functions for modulated renewal processes .
the main distinction between these methods lies in the way they handle the form and integration of the intensity function @xmath46 .
approaches include using kernel - smoothing @xcite , using parametric intensity functions @xcite , using discretized bins within which the intensity is considered constant @xcite , or using a form of rejection sampling called _ thinning _ @xcite that avoids the integration altogether .
the binned time approach is straightforward , and we share its use of gaussian processes for the log intensity function . however , there is an inherent information loss in the piecewise - constant intensity function approximation .
moreover , its computational complexity is cubic with the number of bins in the period of observation . for our data , with events at 1-day or finer time resolution over up to a 15 year observation period ,
this method is prohibitively inefficient . a variant of this approach that uses variable - sized bins @xcite has been applied to medical data @xcite .
this variant is very efficient , but is restricted to a poisson process ( fixed @xmath18 ) , and the inferred intensity functions are neither intended to nor particularly well suited to forming an accurate abstraction over the raw discrete events .
thinning is a clever method , but it is limited by the requirement of a bounded hazard function , which prevents it from being used with bursty gamma processes ( which have a hazard function unbounded at zero ) .
one thinning method has also adopted the use of gaussian processes @xcite , but is much less efficient than our algorithm , with time complexity cubic in the number of events that would occur if the maximum event intensity were constant over the entire observation time span . for event streams with a small dynamic range of intensities , this is not a big issue , but our medical data sequences can have a dynamic range of several orders of magnitude .
our method therefore has efficiency and flexibility advantages over existing methods , and we will demonstrate in the experiments that it also has accuracy advantages .
in these experiments , we will refer to our inference method as the _ direct method _ because it uses direct numerical integration , as opposed to thinning , which avoids computing the integral at all .
we tested the ability of both methods to recover known intensity functions and shape parameters from synthetic data .
we then used the direct method to infer latent intensity functions from streams of clinical events .
.49 .49 our first experiments were with the three parametric intensity functions below , carefully following @xcite and @xcite .
we generated all data using the warping model described in ( section [ sec : mrp ] ) , with shape parameter @xmath69 .
1 . @xmath70 over the interval @xmath71}$ ] , 48 events
@xmath72 on @xmath73 $ ] , 29 events . 3 .
@xmath74 is the piecewise linear curve shown in figure [ fig : parametric - synthetic ] , on the interval @xmath75 $ ] , 230 events .
we express these as normalized intensities @xmath76 , which have units of `` expected number of events per unit time '' , because they are more interpretable than the raw intensities and they are comparable to the previous work done using poisson processes , where @xmath18 . we compared the direct method to thinning on these datasets .
@xcite compared thinning to the kernel smoothing and binned time methods ( all assuming a poisson process ) , and @xcite compared thinning to binned time , assuming a gamma process with constrained @xmath17 .
both found thinning to be at least as accurate as the other methods in most tests .
we computed the rms error of the true vs. the median normalized inferred intensity , the log probability of the data given the model , and the inference run time under 1000 burn - in and 5000 inference mcmc iterations . on these datasets
the direct method was more accurate than thinning for the recovery of both the intensity function and the shape parameter , and more efficient by up to two orders of magnitude ( figure [ fig : parametric - synthetic ] and table [ tab : parametric - synthetic ] ) .
the results for thinning are consistent with those previously reported @xcite .
.performance on synthetic data .
rms : root - mean - squared error ; lp : log probability of data given the model ; rt : run time in seconds .
best results for each measure are bolded . [ cols="<,^,^,^,^,^,^ " , ] additionally , the confidence intervals from the direct method are subjectively more accurate than from the thinning method .
( that is , the 95% confidence intervals from the direct method contain the true function for about 95% of its length in each case ) .
this is particularly important in the case of small numbers of events .
as might be expected , we found the results for @xmath77 to be sensitive to the prior distribution on @xmath22 , given the small amount of evidence available for the inference .
following @xcite and @xcite , we used a log - normal prior with a mode near @xmath78 , tuned slightly for each method to achieve the best results .
we also allowed thinning to use a log - normal prior with appropriate modes for @xmath79 and @xmath80 , to follow precedent in the previous work , although it may have conferred a small advantage to thinning .
we used the weaker exponential prior on those datasets for the direct method .
our next experiments were on synthetic data generated to resemble our medical data .
we tested several configurations over wide ranges of parameters , including some that were not amenable to any known existing approach ( such as the combination of @xmath16 , high dynamic range of intensity , and high ratio of observation period to event resolution , figure [ fig : synthetic ] ) .
the inferred intensities and gamma parameters were consistently accurate .
estimates of the confidence intervals were also accurate .
.49 .49 lastly , we applied the direct method to sequences of billing codes representing clinical events . after obtaining irb approval , we extracted all codes from five patient records with the greatest number of such codes in the deidentified mirror of our institution s electronic medical record .
we arranged the codes from each patient record as streams of events grouped at both the top level of the icd-9 disease hierarchy ( groups of broadly related conditions ) , and at the level of the individual disease .
for the streams of grouped events , we included an event if its associated icd-9 code fell within the range of the given top - level division .
for example , any event with a code in the range [ 390 459.81 ] was considered a _ cardiovascular _ event .
while intensity functions are only strictly additive for poisson processes , we still find the curves of grouped events to be informative .
we inferred intensity functions for each of these event streams ( figures [ fig : clinical ] and [ fig : summary ] ) .
each curve was generated using 2000 burn - in and 2000 inference iterations in about three minutes using unoptimized matlab code on a single desktop cpu .
the results of both types have good clinical face validity .
the set of top - level intensity functions make clear that there is much underlying structure in these originally irregular and asynchronous medical events that can now be investigated with standard learning methods ( figure [ fig : summary ] ) . for example
, there are obvious dependencies in this patient between congenital , cardiovascular , and pulmonary conditions , and then a dependence of those with a genitourinary condition emerges around year 3 .
such structure can be investigated at both the patient and the population level using these abstractions .
we have made two contributions with this paper .
first , we presented a direct numeric method to infer a distribution of continuous intensity functions from a set of episodic , irregular , and discrete events .
this direct method has increased efficiency , flexibility , and accuracy compared to the best prior method .
second , we presented results using the direct method to infer a continuous function density as an abstraction over episodic clinical events , for the purposes of transforming the raw event data into a form more amenable to standard machine learning algorithms .
the clinical interpretation of these intensity functions is that increased intensity represents increased frequency of contact with the healthcare system , which usually means increased _ instability _ of that condition . in some cases ,
it may also mean increased _ severity _ of the condition , but not always . if a condition acutely increases in severity , this represents an instability and will probably generate a contact event .
on the other hand , if a condition is severe but stably
so , it may or may not require high - frequency medical contact . a method to construct similar curves from observations with both a time and a continuous value has been previously reported @xcite , and we have presented here a method to construct them from observations with a time plus a categorical label .
these two data types represent the majority of the information in a patient record ( if we consider words and concepts in narrative text to be categorical variables ) , and opens up many possibilities for finding meaningful patterns in large medical datasets .
the practical motivation for this work is that once we have the continuous function densities , we can use them as inputs to a learning problem in the time domain ( such as identifying trajectories that may be characteristic of a particular disease ) , or by aligning many such curves in time and looking for useful patterns in their cross - sections ( which to our knowledge has not yet been reported ) .
we discovered incidentally that a presentation such as figure [ fig : summary ] appears to be a promising representation for efficiently summarizing a complicated patient s medical history and communicating that broad summary to a clinician .
the presentation could allow drilling - down to the intensity plots of the specific component conditions and then to the raw source data .
( the usual method of manually paging through the often massive chart of a patient to get this information can be a tedious and frustrating process . )
one could also imagine presenting the curves of not the raw icd-9 divisions , but the inferred latent factors underlying them , and drilling down into the rich combinations of test results , medications , narrative text , and discrete billing events that comprise those latent factors .
we plan to investigate these possibilities in future work .
this work was funded by grants from the edward mallinckrodt , jr .
foundation and the national institutes of health 1r21lm011664 - 01 .
clinical data was provided by the vanderbilt synthetic derivative , which is supported by institutional funding and by the vanderbilt ctsa grant ultr000445 .
r. p. adams , i. murray , and d. j. c. mackay .
tractable nonparametric bayesian inference in poisson processes with gaussian process intensities . in _ proceedings of the 26th international conference on machine learning _ , 2009 .
t. a. lasko , j. c. denny , and m. a. levy .
computational phenotype discovery using unsupervised feature learning over noisy , sparse , and irregular clinical data .
_ plos one _ , 80 ( 6):0 e66341 , 2013 .
doi : 10.1371/journal.pone.0066341 .
p. a. w. lewis .
recent results in the statistical analysis of univeraite point processes . in p.
a. w. lewis , editor , _ stochastic point processes ; statistical analysis , theory , and applications_. wiley interscience , 1972 .
p. martinsson , v. rokhlin , and m. tygert . a fast algorithm for the inversion of general toeplitz matrices . _ computers & mathematics with applications _ , 500 ( 56):0 741 752 , 2005 .
doi : http://dx.doi.org/10.1016/j.camwa.2005.03.011 .
i. murray and r. p. adams .
slice sampling covariance hyperparameters of latent gaussian models . in j.
lafferty , c. k. i. williams , r. zemel , j. shawe - taylor , and a. culotta , editors , _ advances in neural information processing systems _ , 2010 .
j. weiss and d. page .
forest - based point process for event prediction from electronic health records . in _
machine learning and knowledge discovery in databases _
, volume 8190 of _ lecture notes in computer science _ , pages 547562 .
springer berlin heidelberg , 2013 .
doi : 10.1007/978 - 3 - 642 - 40994 - 3_35 . | the episodic , irregular and asynchronous nature of medical data render them difficult substrates for standard machine learning algorithms .
we would like to abstract away this difficulty for the class of time - stamped categorical variables ( or _ events _ ) by modeling them as a renewal process and inferring a probability density over continuous , longitudinal , nonparametric intensity functions modulating that process .
several methods exist for inferring such a density over intensity functions , but either their constraints and assumptions prevent their use with our potentially bursty event streams , or their time complexity renders their use intractable on our long - duration observations of high - resolution events , or both .
in this paper we present a new and efficient method for inferring a distribution over intensity functions that uses direct numeric integration and smooth interpolation over gaussian processes .
we demonstrate that our direct method is up to twice as accurate and two orders of magnitude more efficient than the best existing method ( thinning ) .
importantly , the direct method can infer intensity functions over the full range of bursty to memoryless to regular events , which thinning and many other methods can not .
finally , we apply the method to clinical event data and demonstrate the face - validity of the abstraction , which is now amenable to standard learning algorithms .
= 1 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.